Intelligent Digitization and Preservation of Audiovisual Archives

TSAWhitepaper Intelligent Digitization and Preservation of Audiovisual Archives   Índice: Chapter.1. The Strategic Importance of Audiovisual Archives in the Digital Era 5 .1.1. The Dual Value: Cultural Heritage and Commercial Asset 5 .1.2. The Preservation Imperative: Confronting Obsolescence and Degradation 6 .1.3. La obsolescencia de los equipos de reproducción como factor crítico 7 .1.4. Frameworks for Trust: OAIS, PREMIS and Risk Assessment Models 8 Chapter.2. Mastering Digitization and Long-Term Preservation 11 .2.1. Digitization Workflows and Optimal Technical Parameters 11 .2.2. Advanced Digital Restoration and Scalable Quality Control 15 .2.3. Sustainable Storage Ecosystems: LTO, Cloud, Hybrid Architectures and TCO 18 Chapter.3. Artificial Intelligence: Transforming Audiovisual Archive Management 22 .3.1. AI-Powered Content Analysis: Recognition, Transcription, and Segmentation 22 .3.2. Automating Understanding: AI for Metadata Generation and Enrichment 23 .3.3. Realizing the Potential of AI: Applications, Performance, and Current Limitations (Including Case Studies) 25 Chapter.4. Robust Metadata Strategies for Search and Management 28 .4.1. 4.1 Fundamental Audiovisual Metadata Standards: PBCore, EBUCore, MPEG-7 28 .4.2. Achieving Interoperability and Managing Rights Metadata 30 .4.3. Implementing Effective Metadata Validation and Enrichment Workflows 32 Chapter.5. Unlocking Value: Access, Distribution, and Monetization 34 .5.1. Enabling Access: MAM/DAM Platforms and Streaming Technologies 34 .5.2. 5.2. Diversifying Revenue: Licensing, FAST, AVOD, and ROI Considerations 36 .5.3. Ensuring Security, Optimizing the User Experience, and Exploring Federated Access 38 Chapter.6. Industry Implementations and Future Horizons 41 .6.1. Learning from the Leaders: Case Studies in Intelligent Archiving 41 .6.2. Emerging Frontiers: Generative AI, Blockchain, Immersive Media Preservation, Federated Collaboration 44 Chapter.7. Strategic Recommendations for Audiovisual Digitization and Preservation Projects 47 .7.1. Conduct a Comprehensive Risk Assessment and Prioritization 47 .7.2. Standardize Workflows and Preservation Formats 47 .7.3. Design a Sustainable and Secure Storage Architecture 48 .7.4. Rely on Appropriate Audiovisual Asset Management Platforms 48 .7.5. Explore the Use of Artificial Intelligence Progressively and in a Controlled Manner 48 .7.6. Define a Comprehensive Metadata and Rights Management Strategy 49 .7.7. Integrate Access and Reuse Considerations from the Outset 49 .7.8. Promote Continuous Training and Ongoing Technology Monitoring 49 Chapter.8. Conclussion 50 Chapter.9. Glossary of Key Terms 51 Introduction Over the past decades, audiovisual archives were perceived as a legacy of the past. Today, however, they have become one of the most fragile—and paradoxically most valuable—assets for many organizations. Recent television successes recovering national historical memory through high‑audience programs clearly demonstrate this reality. In broadcasters, production companies, public institutions, historical archives and large corporations, millions of hours of audiovisual content remain stored on media that are no longer manufactured, depend on technologies that are disappearing, or simply cannot be reliably played back. The risk is not theoretical. It is cumulative, silent, and in many cases irreversible. Unlike other digital assets, an audiovisual archive that cannot be played back effectively ceases to exist, even if it still occupies physical or logical storage space. Added to this structural risk is a second factor: a growing portion of the value of audiovisual archives remains latent. Content with cultural, informational or commercial potential remains inaccessible, difficult to locate, or impossible to reuse efficiently—not because it lacks value, but because it has not been prepared to operate in a digital environment. In this context, inaction is no longer a neutral position. Delaying decisions means accepting a progressive loss of heritage, reuse opportunities and future responsiveness. Acting without a clear strategy—simply digitizing without criteria, standards or a long‑term vision—introduces different but equally costly risks. For the first time, audiovisual preservation clearly enters the realm of strategic decision‑making. Not as an isolated technical task, but as an investment that conditions an organization’s ability to preserve its memory, exploit its assets and adapt to new access, distribution and monetization models. This White Paper is grounded in that reality: first understanding the risk, then defining the correct approach, and finally identifying the tools that enable audiovisual archives to be transformed into sustainable digital assets.   Executive Summary Audiovisual archives concentrate an essential part of collective memory while also representing a growing volume of latent economic value. However, a significant portion of this heritage is currently in a critical situation: media that degrade irreversibly, playback technologies that are disappearing, and content that—despite its value—remains inaccessible or unusable. In this context, digitization can no longer be understood as a one‑off technical task, but as a strategic decision. Inaction implies accepting the definitive loss of cultural, operational and commercial assets. Acting without an appropriate strategy entails equivalent risks, from premature obsolescence to investments that are difficult to sustain over time. This White Paper analyzes how to address audiovisual digitization and preservation from a comprehensive perspective, combining technical, organizational and economic criteria. Throughout the document, it examines the real risks associated with media degradation and technological obsolescence, as well as best practices consolidated across the industry to ensure long‑term preservation. The report progressively explores the key components of a sustainable audiovisual preservation strategy: rigorous digitization workflows, open standards, secure and efficient storage architectures, advanced metadata management, and the emerging role of artificial intelligence as an accelerator for content analysis and reuse. The objective is not to propose a single solution, but to offer a reference framework that enables organizations to make informed decisions aligned with their long‑term objectives and prepared for an increasingly digital, distributed and demanding audiovisual environment.  Chapter.1. The Strategic Importance of Audiovisual Archives in the Digital Era Audiovisual archives have transcended their traditional role as mere repositories of the past to become dynamic assets of immense cultural, historical and increasingly commercial value. In this context, it is essential to understand the dual value of audiovisual archives and the factors that condition their preservation and value creation within today’s media environment. .1.1. The Dual Value: Cultural Heritage and Commercial Asset Audiovisual archives are irreplaceable custodians of collective memory. They capture not only historical events, but also cultural traditions, social practices, and the “emotional essence of different eras.” They function as vital testimonies of the past, offering first-hand narratives through moving images and sound, capturing nuances such as facial expressions and vocal inflections that text cannot convey. For communities and nations, these archives are fundamental to identity, mutual understanding, and historical reflection. Even government agencies formally recognize audiovisual materials as official records with unique informational value that reflect broad spectra of national life and are crucial for research. Beyond their intrinsic value as heritage, these archives represent a significant commercial asset. They have been compared to raw materials such as “oil” or “iron” which, once refined (digitized and managed), fuel the information, communication, and creative industries. Content can be licensed for use in new productions, advertising, documentaries, or educational platforms, generating direct revenue streams. Digitization is the key catalyst that unlocks this potential, enabling broader access, easier reuse, and new business models. The traditional view of archives as cost centers dedicated exclusively to cultural preservation is obsolete. In the digital domain, investment in preservation (digitization, cataloguing, managed storage) directly enables both the cultural mission and commercial opportunities, building a compelling business case. The COVID-19 pandemic also accelerated the shift toward digital cultural consumption, underscoring the growing demand for—and potential value of—online-accessible audiovisual archives. Although economic models for fully valuing digital culture are still developing, evidence suggests positive social value derived from online access and from ensuring that cultural heritage remains digitally accessible for present and future generations. This reinforces the strategic importance of digitization not only as a preservation measure, but as an investment in future relevance and value. .1.2. The Preservation Imperative: Confronting Obsolescence and Degradation The need to preserve audiovisual archives is urgent and complex, driven by two interconnected threats: the physical and chemical degradation of original carriers and the technological obsolescence of formats and equipment. • Physical / Chemical Degradation: Audiovisual media are inherently fragile. o Magnetic Tapes (Audio and Video): They are susceptible to binder hydrolysis—a chemical process exacerbated by humidity and temperature—in which the binder that holds the magnetic particles to the tape base breaks down. This causes the phenomenon known as sticky shed syndrome, where the tape becomes sticky, sheds residue that clogs playback heads, increases friction, and in extreme cases can halt tape transport. Loss of lubricant over time—due to use, evaporation, or chemical degradation—also increases friction and wear. The stability of magnetic pigments (iron oxide, chromium dioxide, metal particles) varies as well, affecting long-term signal retention. Improper storage (heat, humidity, light, contaminants) dramatically accelerates these processes. o Cinematographic Film: Historical film bases present severe stability problems. Cellulose nitrate (used until ~1950) is chemically unstable and highly flammable, degrading over time. Cellulose acetate, its “safe” replacement, also degrades through an autocatalytic process known as vinegar syndrome, releasing acetic acid (a vinegar smell), becoming brittle, shrinking, and eventually becoming unusable. In practice, there is not much difference between nitrate and acetate in terms of deterioration. In addition, the organic dyes used in color films are inherently unstable and fade over time, especially under warm and humid conditions. o Optical and Early Digital Media: Although often perceived as more stable, media such as CDs and DVDs also have limited lifespans and are vulnerable to physical damage (scratches) and chemical degradation of their layers. • Technological Obsolescence: a pervasive threat affecting all formats, analog and digital. o Hardware: Playback equipment for legacy formats (VTRs for formats such as Umatic, Betacam, 1-inch C; film projectors for different gauges; open-reel or cassette audio players; older optical disc drives) becomes increasingly scarce, expensive to maintain, and difficult to repair due to lack of parts and technical expertise. o Software y File formats: Digital file formats—both containers and codecs—also evolve. New software versions may stop supporting older formats (planned or accidental obsolescence), making data inaccessible. This affects both files digitized from analog carriers and “born-digital” materials in older formats. The proliferation of multiple formats and versions within a collection also complicates management. These two threats—degradation and obsolescence—act synergistically. As a physical carrier degrades, the window of opportunity to digitize it using increasingly scarce playback equipment closes. Conversely, the lack of functional equipment can render a physically stable carrier inaccessible. This dynamic creates a critical urgency for proactive digitization. It is crucial to understand that digitization is not an end point, but the beginning of an ongoing digital preservation lifecycle. Digital files themselves face risks such as data corruption (bit rot), file format obsolescence, and storage media failure. Therefore, active and continuous management is required—including monitoring, format migration, and storage media refresh—to ensure long-term accessibility. Planning and budgeting must extend beyond the initial digitization project to cover this ongoing digital curation. .1.3. La obsolescencia de los equipos de reproducción como factor crítico Digitizing magnetic-tape-based audiovisual archives faces a critical factor that goes beyond the physical deterioration of the carriers: the progressive disappearance of the equipment required for playback. Many video formats—both professional and consumer—depend on specific VTRs that are no longer manufactured and whose operational availability is rapidly declining. The situation is exacerbated by the scarcity of essential components, especially playback heads. These elements, fundamental to reading the recorded signal, are no longer produced and can only be obtained through reuse or refurbishment of out-of-service equipment. The closure of the last specialized suppliers in this area marks a turning point: once existing spare parts are exhausted, numerous formats will become definitively inaccessible. In addition to this technical obsolescence, there is a loss of specialized knowledge. Operating, maintaining, and aligning these systems requires expertise that is increasingly rare, as many professionals trained on these technologies are retiring without sufficient replacement. As a result, even equipment that still functions does so in a context of high operational fragility and rising costs. The combined impact of these factors drastically reduces the time window available for digitization. For certain formats, this window is already measured in just a few years—not only because tapes are degrading, but because it is becoming practically impossible to play them back. Delaying digitization means accepting the real risk that content cannot be recovered, even if the carriers are physically preserved. From a strategic perspective, this reinforces the need to approach digitization as a race against time, in which technical capacity, availability of specialized infrastructure, and large-scale planning are decisive to avoid the irreversible loss of audiovisual heritage. Once the magnitude of the risk is understood, the question is no longer whether action is necessary, but how to do it correctly. Sector experience shows that audiovisual preservation is only sustainable when it is addressed as a structured program, supported by clear reference frameworks, defined policies, and coherent technical decisions. Without this approach, digitization risks becoming a costly effort that is difficult to sustain over time. .1.4. Frameworks for Trust: OAIS, PREMIS and Risk Assessment Models To systematically address the challenges of digital preservation, the archival community has developed conceptual frameworks, metadata standards, and risk assessment methodologies. These provide a solid foundation for building trustworthy digital repositories and sustainable preservation strategies. • OAIS Reference Model (Open Archival Information System – ISO 14721): OAIS is the fundamental and most widely adopted conceptual framework for digital archives. It defines an OAIS as an organization of people and systems responsible for preserving information and making it available to a defined “Designated Community.” It establishes six key mandatory responsibilities for a trustworthy archive: 1. Negotiate for and accept appropriate information from producers. 2. Obtain sufficient control for long-term preservation. 3. Determine the scope of the Designated Communit. 4. Ensure the information is independently understandable by the Designated Community. 5. Follow documented policies and procedures to preserve against contingencies and avoid ad hoc deletions. 6. Make the information available and allow the distribution of authenticated copies. The model also defines six essential functional entities that structure archive operations: Ingest (receiving and preparing materials), Archival Storage (long-term custody and maintenance of information packages), Data Management (handling descriptive and administrative metadata), Administration (overall operations and policy management), Preservation Planning (monitoring risks and developing strategies such as migration or emulation), and Access (interfaces that enable users to find and obtain information). OAIS provides a common language and a logical structure that are indispensable for planning, implementing, and comparing digital preservation systems. • PREMIS Data Dictionary (Preservation Metadata: Implementation Strategies – ISO 22957): PREMIS is the de facto international standard for preservation metadata. It defines a core set of “semantic units” (metadata elements) necessary to support long-term preservation functions. PREMIS is structured around a data model with four main entities: 1. Object: What is being preserved. This can be an Intellectual Entity (the conceptual work), a Representation (a set of files embodying the intellectual entity), a File (a named sequence of bytes), or a Bitstream (data within a file). 2. Event: Significant actions that occur during the lifecycle of a digital object within the repository (e.g., creation, ingest, format validation, migration, fixity checks). 3. Agent: People, organizations, or software involved in preservation events. 4. Rights: Information about permissions and restrictions specifically relevant to preservation actions (e.g., permission to migrate formats, permission to create copies). PREMIS is crucial for documenting provenance (custodial history and changes), authenticity, technical characteristics, and the context of digital objects—information that is indispensable to ensure future usability. • Risk Assessment Models and Tools: Given the scale of archives and limited resources, risk assessment is essential for prioritizing actions and allocating funds effectively. There are various methodologies: o Format-based: Assess the inherent risk of specific formats based on physical fragility and technological obsolescence. Tools such as FACET (Field Audio Collection Evaluation Tool), AVPRAPPS (Audiovisual Preservation Readiness Assessment Project Planning System), and the NARA framework with its Risk Matrix use scoring systems based on these characteristics. o Institutional capability-based: Models such as the NDSA Levels of Digital Preservation (National Digital Stewardship Alliance) and DPC RAM (Digital Preservation Coalition Rapid Assessment Model) evaluate the maturity of an institution’s technical and organizational capabilities for digital preservation. o Integrated models: Aim to quantify overall risk by combining format factors, organizational capacity, and potential impacts. Examples include DiAGRAM (Digital Archives Graphical Risk Assessment Model), PRISM (Preservation Risk Information System Model), and BPRisk (focused on workflow risks). These three pillars—OAIS, PREMIS, and risk assessment—are intrinsically connected and form the foundation of a coherent digital preservation strategy. OAIS defines the functional structure and responsibilities (what to do). PREMIS specifies the information needed to document and manage objects and processes within that structure (what to record). Risk assessment helps determine where and when to apply resources within the OAIS framework, prioritizing the most vulnerable or valuable assets, based in part on technical information (PREMIS metadata) about formats. Together, they provide a systematic approach for informed decision-making in long-term digital preservation. With this strategic framework defined, the next step is to translate these principles into practice. Digitization and long-term preservation are not abstract processes: they require well-defined workflows, appropriate technical parameters, and control mechanisms that ensure the quality and integrity of digital assets from the very beginning.  Chapter.2. Mastering Digitization and Long-Term Preservation The transition of audiovisual archives from the analog domain to the digital domain, and the ongoing management of the resulting digital assets, requires the implementation of rigorous workflows, the adoption of optimal technical parameters, and the development of sustainable storage and quality control strategies. .2.1. Digitization Workflows and Optimal Technical Parameters An effective digitization workflow is a planned and standardized process that ensures the creation of high-quality digital files and their proper integration into the preservation ecosystem. • Workflow Stages: Although details may vary, a typical workflow includes: 1. Planning and Prioritization: Selection of materials based on risk criteria (degradation, obsolescence), content value (uniqueness, historical or cultural importance), usage value (demand, alignment with institutional mission), and digitization cost. Identification and removal of ROT data (Redundant, Obsolete, Trivial). 2. Material Preparation: Detailed item-level inventory, assignment of unique identifiers, physical inspection to assess condition (e.g. mold, vinegar syndrome, physical damage), cleaning and minor repairs where necessary and feasible. 3. Capture / Digitization: Transfer of audiovisual content into digital formats using calibrated equipment and appropriate technical settings. 4. Quality Control (QC): Verification of the generated digital files to ensure they meet technical specifications and faithfully represent the original material (see Section 2.2). 5. Metadata Creation: Generation or capture of technical, descriptive and preservation metadata (see Section 4). 6. Ingest: Transfer of validated digital files and their associated metadata into the archive storage or management system (e.g. MAM/DAM, digital repository). • Key Technical Parameters (for Preservation Masters): Selecting appropriate technical parameters during capture is critical to creating a high-fidelity digital master that can serve as a long-term surrogate for the original analog material. Recommendations from organizations such as FADGI (Federal Agencies Digitization Guidelines Initiative), IASA (International Association of Sound and Audiovisual Archives), EBU (European Broadcasting Union), and SMPTE (Society of Motion Picture and Television Engineers) guide best practices: o Resolution (Video/Film): Must be sufficient to capture the detail of the original. Common examples include 4K (4096 pixels wide) for 35 mm film and 2K (2048 pixels wide) for 16 mm / 8 mm film. For SD video, native resolution (e.g. 720×480 NTSC) or appropriate scaling may be used. FADGI employs a star system (1–4), with four stars representing preservation-level quality. o Bit Depth (Video/Film/Audio): Determines dynamic range and color or tonal precision. High bit depths are recommended: 24 bits per sample for audio, and 10 bits or even 16 bits per color component for video and film. o Sampling Rate (Audio): Captures the frequency range. For archival masters, 96 kHz or even 192 kHz is recommended. 48 kHz may be acceptable for certain materials, while 44.1 kHz (CD quality) is generally considered insufficient for preservation. o Compression: Preservation masters should use uncompressed encoding or lossless compression. Lossy compression irreversibly discards information and is suitable only for access copies. o Color Space / Encoding: For video, standards such as BT.709 or BT.2020 are used, with encodings such as RGB or YUV (e.g. 4:2:2 for broadcast-quality video). For film, linear or logarithmic color spaces are commonly employed. o Frame Rate (Video / Film): Debe coincidir con la del material original. • Preservation File Formats: he choice of container format and codec is critical. Preferred formats are open, well documented, widely adopted, and suitable for long-term preservation. o Video/Film:  FFV1 (códec lossless) in Matroska (MKV) (contenedor): An open-source option increasingly recommended by archival institutions for its efficient lossless compression and the flexibility of the MKV container.  JPEG 2000 (lossless o lossy) in MXF (wrapper): Common in broadcast and archival environments, particularly for scanned film. MXF is a robust professional container.  Uncompressed Video (e.g. v210, 10-bit YUV) in QuickTime (MOV) or AVI: Provides maximum fidelity but generates very large files. MOV is proprietary to Apple; AVI has limitations in file size and metadata handling.  Image Sequences (DPX or TIFF): The de facto standard for high-quality film scanning. Each frame is stored as an individual image file. OpenEXR is another option for high dynamic range workflows.  Apple ProRes (4444 o 422 HQ): High-quality codecs often used as intermediate or mezzanine formats, but proprietary to Apple. Avid DNxHD / DNxHR provide comparable alternatives. o Audio:  BWF (Broadcast Wave Format) with LPCM (audio without compression): An extension of the standard WAV format that allows embedded metadata (defined by EBU Tech 3285). It is the primary recommendation of IASA. File size limitations (4 GB) can be overcome using RF64 / MBWF.  WAV (Waveform Audio File Format) with LPCM: A widely supported uncompressed standard.  AIFF (Audio Interchange File Format) con LPCM: Apple’s equivalent of WAV, also uncompressed.  FLAC (Free Lossless Audio Codec): An open-source lossless compression format that reduces file size without compromising audio quality. • Access Formats: For distribution and online use, derivative copies are created in lossy compressed formats optimized for streaming and reduced file size. Common examples: o Video: H.264 (AVC) or H.265 (HEVC) in MP4. o Audio: MP3 or AAC. The fundamental principle underlying these recommendations is the separation between preservation masters and access copies. Preservation masters are created at the highest possible fidelity and stored securely for long-term preservation, serving as the source for future derivatives. Access copies, smaller and lower quality, are generated from the master to facilitate distribution, streaming, and everyday use. This two-tier strategy is essential for sustainable digital archive management. Table 1. Recommended Formats for Audiovisual Preservation Masters Original Content Type Recommended Container Recommended Video Codec Recommended Audio Codec Suggested Key Parameters Film (35 mm) MKV or MXF or Image Sequence FFV1 (in MKV) or JPEG 2000 (lossless, in MXF) or DPX / TIFF / OpenEXR (sequence) LPCM or FLAC (if sound is present) 4K (or higher), 10/16-bit, Linear or Log color space, original frame rate Film (16 mm / 8 mm) MKV or MXF or Image Sequence FFV1 (in MKV) or JPEG 2000 (lossless, in MXF) or DPX / TIFF (sequence) LPCM or FLAC (if sound is present) 2K (or higher), 10-bit, Linear or Log color space, original frame rate Analog Video (Broadcast – e.g. 1”, Betacam SP) MKV or MXF or MOV FFV1 (in MKV) or Uncompressed (10-bit YUV, e.g. v210 in MOV) or JPEG 2000 (lossless, in MXF) LPCM Native resolution / HD, 10-bit, 4:2:2 YUV, original frame rate Analog Video (Consumer – e.g. VHS, Hi8) MKV or MOV FFV1 (in MKV) or Uncompressed (8/10-bit YUV) LPCM Native resolution / SD, 8/10-bit, original frame rate Analog Audio (Open Reel, Cassette) BWF or WAV or FLAC N/A LPCM (in BWF/WAV) or FLAC 96 kHz (or 192 kHz), 24-bit Digital Audio (DAT, CD) BWF or WAV or FLAC N/A LPCM (in BWF/WAV) or FLAC Original sample rate and bit depth (if ≥ 44.1 kHz / 16-bit), or higher (e.g. 96 kHz / 24-bit) (Note: This table provides a general synthesis. Exact specifications may vary depending on the specific material and institutional objectives. Consultation of FADGI and IASA guidelines is recommended). .2.2. Advanced Digital Restoration and Scalable Quality Control Once material has been digitized, restoration processes are often required to correct defects present in the original media or artifacts introduced during capture, alongside rigorous quality control to ensure the long-term viability of the resulting digital assets. • Digital Restoration Techniques: The objective is to improve visual and/or audio quality while respecting the integrity of the original content. o Image Restoration (Video / Film):  Stabilization: Correction of image vibration or frame instability.  Cleaning: Digital removal of dust, scratches, stains, and hairs (dust busting, scratch removal). This can be performed manually (frame by frame, labor-intensive) or through automated or semi-automated specialized software (e.g. Digital Phoenix).  Color Correction: Adjustment of luminance, chrominance, and white balance to restore original colors or compensate for fading.  Noise / Grain Reduction: Attenuation of electronic noise or film grain, taking care not to remove fine detail.  Damage Repair: Reconstruction of damaged or missing frames, which may involve interpolation or the use of generative AI, with associated authenticity considerations. o Audio Restoration:  Noise Reduction: Removal of tape hiss, electrical hum, clicks, and pops.  Equalization: Correction of tonal imbalances.  Dynamic Restoration: Expansion or compression to restore original dynamic range. o Treatment of Specific Media: For magnetic tapes affected by sticky shed syndrome, controlled “baking” may temporarily enable successful transfer, but must be performed with caution and followed immediately by digitization. Careful physical cleaning of tapes and discs is essential prior to capture. o Generative AI in Restoration: Emerging AI technologies can “invent” details to increase apparent resolution (upscaling) or colorize black-and-white films. While they may improve viewing experience, they introduce artifacts (the “AI look”) and raise serious concerns about historical authenticity. Their use in strictly archival contexts requires extreme caution and transparent documentation. This tension between aesthetic enhancement and historical fidelity is a key dilemma: advanced restoration can make content more attractive, but may distance it from the original artifact. • Scalable Quality Control (QC): Quality control is indispensable for validating technical quality and content integrity, particularly in large-scale digitization projects. A scalable approach combines automated and manual methods: o Automated QC (100% of files):  Fixity Verification: Calculation and comparison of checksums (e.g. MD5, SHA-256) to ensure files have not been altered during transfer or storage. Tools such as BagIt can assist.  Format Validation: Verification that files conform to their declared format specifications (e.g. using JHOVE, MediaInfo, MediaConch), detecting corruption or malformed files.  Technical Specification Checks: Automated verification of key parameters (resolution, bitrate, codec, sampling rate, etc.) against project requirements.  Metadata Validation: Verification of presence, format, and validity of embedded or associated metadata (e.g. using BWF MetaEdit or schema validators). o Manual QC (Representative sample, e.g. 10–15%): Performed by qualified staff to detect subjective or contextual issues that machines cannot identify.  Critical Listening / Viewing: Detection of visual artifacts (glitches, macroblocking, color errors), audio issues (distortion, clicks, audio-video desynchronization), or incomplete or incorrect content.  Metadata Review: Verification of accuracy and relevance of descriptive metadata and digitization engineer notes.  Assistance Tools: Software such as QCTools (video waveform analysis) and Sonic Visualizer (audio waveform analysis) can support detailed inspection. o Integration into the Workflow: QC must be systematically integrated into workflows after capture and before final ingest. A robust QC strategy mitigates risk, protects digitization investment, validates asset integrity over time, and underpins trust in the digital repository. Implementing robust quality control (QC) is a fundamental risk-mitigation strategy. It ensures that investment in digitization is not wasted on unusable or corrupted files, validates the integrity of digital assets over time, and underpins trust in the digital repository as a reliable custodian of heritage. Without effective QC, an archive may unknowingly store corrupted, unusable, or incomplete data, compromising its preservation mission.   .2.3. Sustainable Storage Ecosystems: LTO, Cloud, Hybrid Architectures and TCO Long-term digital preservation requires robust, scalable, secure, and economically sustainable storage solutions. The selection of storage technology and architecture is therefore a key strategic decision. • Main storage technologies: o Hard Disk Storage (HDD – Spinning Disk): Commonly used for online or nearline storage (NAS/SAN environments), providing fast response times, allowing inmediate access and active monitoring. However, it is generally the most expensive option in terms of hardware, energy consumption (always powered on), and maintenance. SSDs offer higher performance but are typically cost-prohibitive for large-scale archival storage. o Magnetic Tape (LTO – Linear Tape-Open): The dominant technology for long-term archival storage and backups in the media sector, especially for cold data to which access is not very frequent:  Advantages: High capacity per cartridge (LTO-9: 18 TB native / 45 TB compressed), low cost per TB, excellent energy efficiency (power is only consumed during read/write operations), long media lifespan when properly stored, and high security by enabling an air gap (physical disconnection from the network) that protects against ransomware and other cyber threats. The native capacity of LTO-10 ranges between 30 TB and 40 TB, depending on the cartridge type within the standard and on each manufacturer’s implementation, while in all cases maintaining compatibility with LTO-10 drives.  Roadmap: The LTO consortium maintains a roadmap for future generations with increased capacity and speed (currently planned up to Generation 14).  Compatibility: Traditionally, LTO drives can read two previous generations and write one previous generation (up to LTO-7). LTO-8 and LTO-9 can only read/write one previous generation. LTO-10 drives can write to LTO-9 cartridges and read LTO-9 and LTO-8 cartridges, following the standard policy of the LTO specification. This limited compatibility requires planning for periodic migrations.  Features: Support for LTFS (Linear Tape File System), that grantes access to files in tape as if they were in a disc, AES-256 encryption (from LTO4), and WORM (Write Once, Read Many) options for regulatory compliance.  Use: Typically implemented in robotic tape libraries for nearline storage or offline archives requiring manual intervention.. o Cloud Storage: Offered by providers such as Amazon Web Services (AWS), Microsoft Azure, Google Cloud Platform (GCP), or TSAmediaHUB, it provides on-demand scalability, access from anywhere, and often built-in geographic redundancy.  Service Tiers:There are different storage tiers, ranging from “hot” (frequent access, higher cost) to “cold” or “archive” (infrequent access, lower storage cost but higher latency and/or retrieval cost), such as AWS S3 Glacier (including Instant Retrieval, Flexible Retrieval, Deep Archive), Azure Blob Storage (Hot, Cool, Archive tiers), and Google Cloud Storage (Standard, Nearline, Coldline, Archive).  Provider Comparison: AWS is the market leader, with the broadest service offering and the largest global infrastructure. Azure stands out for its integration with the Microsoft ecosystem and its hybrid capabilities. GCP is strong in data analytics, AI/ML, and Kubernetes, and often offers competitive pricing.  Considerations: Ongoing costs (monthly storage, data transfer fees or “egress fees” when retrieving data from the cloud), access and retrieval speeds (especially from archive tiers), security (shared responsibility model), regulatory compliance, exit policies (how to retrieve all data if changing providers), and potential vendor lock-in are critical factors to evaluate.. • Hybrid Architectures: The combination of local storage (on-premise, using disks and/or tapes) and cloud storage is an increasingly common and often optimal strategy. It allows organizations to leverage the strengths of each technology: for example, maintaining a master copy on local LTO tape (air-gap security, low long-term cost), a working copy on local disk (fast access), and a backup or disaster recovery (DR) copy in the cloud (geographic redundancy, scalability). This approach makes it possible to balance cost, performance, security, and resilience according to the specific needs of the archive. • Active Storage Management: Regardless of the chosen technology, digital storage is not passive. It requires continuous management, including: o Monitoring: Ongoing supervision of the condition of disks and tapes. o Integrity Checks (Fixity Checks): Regular verification of checksums to detect silent corruption (“bit rot”). This includes data scrubbing (correction of detected errors using redundant copies). o Replication: Maintaining multiple copies of the data (ideally three or more). o Geographic Redundancy: Storing copies in different physical locations to protect against local disasters. o Media Refresh: Periodically replacing storage media (disks, tapes) before they reach end of life or become obsolete (e.g. migrating from LTO-7 to LTO-9). o Format Migration: Planning and executing the migration of files to newer or more sustainable formats if current formats become obsolete or high-risk. • Total Cost of Ownership (TCO): TCO is an essential framework for evaluating and comparing different storage solutions over their lifecycle. It goes beyond the initial acquisition cost (CapEx) and includes all operational costs (OpEx): o TCO Components: Hardware (servers, tape libraries, drives), software (operating systems, storage/archive management software), media (disks, LTO cartridges), energy consumption, cooling, physical data center space, hardware/software maintenance, administrative staff, migration costs (data and format migration), and, in the case of cloud storage, subscription fees, storage charges, data transfer costs (ingress/egress), and support. Indirect or risk-related costs must also be considered, such as the impact of downtime or data loss. o Models and Calculators: Generic models (e.g. SNIA) and specific tools (e.g. Fujifilm’s calculator comparing tape/disk/cloud, and calculators from providers such as Scale Computing, Commercetools, and CTERA) are available to support TCO analysis. Academic studies have also proposed OAIS-based cost frameworks. o Strategic Importance: TCO analysis enables informed investment decisions, realistic comparison of alternatives, budget justification, and long-term financial sustainability planning for the digital archive. Decisions regarding storage must be based on a careful TCO analysis tailored to the specific context of each organization. While LTO tape often offers a lower TCO for long-term archiving of large volumes of cold data due to its low cost per TB and energy efficiency, other factors are critical. Required access frequency, acceptable recovery times, scalability needs, specific security requirements (e.g. the need for an air gap), and TSA’s existing infrastructure and expertise will significantly influence the optimal choice for each client between tape, cloud, disk, or—most likely—a carefully designed hybrid architecture. A generic TCO analysis is insufficient; a customized evaluation is required. Table 2. Comparison of Storage Options for Digital Archives Feature LTO Tape Library (On-Premise) Cloud Storage (Archive Tier) HDD Disk Array (On-Premise) Capacity Density Very High Virtually Unlimited (Scalable) Moderate to High Media Cost (per TB) Very Low Low (storage), Variable (retrieval / egress) Moderate Energy Consumption Low (only during read/write) Variable (depends on provider and tier) High (always powered on) Access Speed / Latency Low (minutes/hours to mount tape) Very Low (hours or days for deep archive retrieval) High (milliseconds) Scalability Moderate (limited by physical library) Very High (on demand) Moderate (requires adding hardware) Security (Air Gap) Possible (offline) No (always connected) No (always connected) Management / Maintenance Moderate (tape handling, migrations) Low (managed by provider) High (hardware, OS, patch management) Typical Archive Use Case Long-term preservation copies, offline backup / disaster recovery Offsite backup, deep archive, disaster recovery Nearline storage for more frequent access, cache Estimated TCO Profile (Long Term, Cold Data) Low Low to Moderate (depends on access / egress) High (Note: TCO profiles are relative and strongly dependent on usage patterns, data volumes, local energy costs, and specific cloud provider pricing). Once the technical foundations of preservation are secured, a new challenge emerges: scale. As archives grow, manually understanding, describing, and managing large volumes of content becomes unfeasible. At this point, artificial intelligence begins to play a key role—not as a replacement for human judgment, but as an accelerator in the analysis, cataloguing, and reuse of audiovisual content.  Chapter.3. Artificial Intelligence: Transforming Audiovisual Archive Management Artificial Intelligence (AI), and in particular machine learning (ML) and deep learning (DL) techniques, are emerging as powerful tools to address some of the most significant challenges in managing large-scale audiovisual archives, especially in content analysis and metadata generation. .3.1. AI-Powered Content Analysis: Recognition, Transcription, and Segmentation AI’s ability to process and “understand” audiovisual data opens new avenues for extracting meaningful information directly from the content itself. • Key Applications: o Automatic Speech Recognition (ASR): Converts spoken language in recordings into written text. This is fundamental for making spoken content indexable and searchable, dramatically improving accessibility and discoverability, particularly for news archives, interviews, or oral history collections. Accuracy may be affected by audio quality, background noise, accents, domain-specific terminology, and disfluent speech (common in historical recordings or among speakers with speech disorders). Models such as OpenAI’s Whisper exemplify advanced ASR technology. o Visual Recognition:  Object Detection and Recognition: Identifies objects within video frames (e.g. cars, buildings, animals).  Facial Recognition: Detects and potentially identifies faces of known individuals, raising ethical and privacy considerations.  Optical Character Recognition (OCR) in Video: Extracts on-screen text (e.g. signs, credits, news headlines).  Scene and Activity Recognition: Classifies scene types (e.g. interior, exterior, studio) or identifies actions taking place. o Análisis de Audio:  Sound Classification: Distinguishes between speech, music, silence, ambient noise, or specific sounds (e.g. applause, sirens).  Speaker Identification (Diarization): Determines who is speaking and when in recordings with multiple participants.  Music Analysis: Identification of genre, tempo, instruments, or even specific melodies.  Sentiment / Emotion Analysis: Attempts to infer emotional states from voice or facial expressions (still under development and with notable limitations). o Video Segmentation: Automatically divides long videos into smaller, meaningful units such as scenes or thematic segments (e.g. separating individual news stories within a newscast, identifying commercial breaks). This facilitates navigation and granular access to content. o Multimodal Analysis: Combines information from different modalities (audio, video, text) to achieve richer understanding. For example, Audio-Visual Speech Recognition (AVSR) uses lip movements alongside audio to improve ASR accuracy in noisy environments. Another example is associating speech transcripts with identified speaker faces. • Underlying Technologies: Primarily deep neural networks, including Convolutional Neural Networks (CNNs) for image analysis (e.g. ResNet), Transformers for sequence processing (e.g. ViViT for video, AST for audio, large language models for text), and task-specific architectures for ASR or diarization.. The most immediate and tangible impact of AI technologies in audiovisual archives is the dramatic improvement in content accessibility and searchability. By automating speech transcription and the identification of key visual and audio elements, AI transforms vast collections that were previously opaque and difficult to navigate into indexed, searchable resources. This enables users to locate specific information far more efficiently than through traditional manual cataloguing or linear viewing. .3.2. Automating Understanding: AI for Metadata Generation and Enrichment One of the major bottlenecks in audiovisual archive management is the creation of detailed descriptive metadata—a process that has traditionally been manual, slow, and costly. AI offers the possibility of automating or semi-automating much of this work. • Automatic Metadata Generation: The outputs of AI-based content analysis (ASR transcripts, object and face labels, scene types, speaker identification, video segments) can be directly converted into structured metadata. For example: o Transcripts can serve as full textual descriptions or be used to automatically extract keywords. o Object and face labels can populate subject or personal name fields. o Segmentation data can generate structural metadata (e.g. scene start/end points). o OCR-extracted text can capture titles or credits. • Metadata Enrichment: AI can go beyond simple extraction to enrich existing metadata: o Entity Linking: Identifying names of people, places, and organizations in transcripts or descriptions and linking them to unique identifiers in external databases (e.g. Wikidata, VIAF, GeoNames), adding context and enabling semantic search. o Summary Generation: Using language models to create concise summaries of video or audio content. o Topic / Category Assignment: Automatically classifying content into predefined categories or identifying emerging themes. • Integration into Workflows: AI tools can be integrated into digitization and ingest workflows, operating fully automatically or in a “human-in-the-loop” mode, where AI suggests metadata that is then reviewed, corrected, and validated by human cataloguers. Platforms such as the Audiovisual Metadata Platform (AMP) explicitly aim to combine automated mechanisms with human labor in recursive workflows. • Domain Adaptation: The effectiveness of AI models often depends on how well they are adapted to the specific content domain (e.g. news, fiction, sports, historical archives). It may be necessary to train or fine-tune models with domain-specific data to improve accuracy and relevance. While the prospect of large-scale automated metadata generation is highly attractive, it is essential to recognize that current AI systems, although powerful, are not infallible. Achieving high-precision, semantically rich metadata—especially for complex, historical, or culturally nuanced content—still typically requires human intervention. A collaborative human-AI approach, where AI performs large-scale extraction and suggestion while human experts validate, correct, and contextualize the results, appears to be the most realistic and effective strategy today. Fully automated, perfect metadata generation remains more an aspiration than an operational reality. .3.3. Realizing the Potential of AI: Applications, Performance, and Current Limitations (Including Case Studies) The practical application of AI in audiovisual archives is expanding, with research projects and early implementations demonstrating both its potential and its inherent challenges. • Case Studies and Examples o Vanderbilt Television News Archive (VTNA): Used ASR and other AI tools to generate transcripts and enhance metadata for its extensive television news archive, with the primary goal of eliminating cataloguing backlogs and making the entire collection accessible to researchers. o News Segmentation Research: A study compared several deep-learning classifiers (ResNet, ViViT, AST, multimodal models) for automatically segmenting news videos into categories such as commercials, stories, and studio scenes, using a custom annotated dataset. Interestingly, image-based classifiers (ResNet) outperformed more complex temporal models in accuracy (84.34%) while requiring fewer computational resources for this specific task. o European Archives (RAI, INA): Institutions such as RAI (Italy) and INA (France) are applying AI within production and deep archives for raw material analysis, content retrieval, and reuse enhancement, often as part of innovation projects or proof-of-concept initiatives. o Assistive AVSR Research: Development of AVSR systems that integrate iconic hand gestures to enrich transcription of non-fluent speech in individuals with speech disorders, demonstrating AI’s multimodal potential. • Performance and Efficiency: AI can dramatically accelerate the processing and analysis of large volumes of audiovisual content compared to manual methods. As shown in the news segmentation study, simpler and more computationally efficient models can sometimes achieve excellent performance for specific tasks. • Current Limitations and Challenges o Accuracy: AI models make errors. The accuracy of ASR, object/face recognition, and classification varies depending on source quality (noise, low resolution, lighting), content diversity (accents, languages, topics), and model suitability for the task. o Bias: AI models learn from training data. If that data contains biases (e.g. imbalanced representation of genders, ethnicities, or topics), the models may perpetuate or amplify those biases, leading to inaccurate or unfair descriptions. o Interpretability and Explainability (XAI): Many deep-learning models function as “black boxes,” making it difficult to understand why specific decisions are made. This lack of transparency can undermine trust, complicate error correction, and raise accountability concerns, especially in sensitive applications. o Computational Cost: Training and running AI models at scale—particularly complex ones—can require significant computational infrastructure and en ergy consumption. o Contextual Understanding: Current AI systems often lack deep understanding of historical, cultural, or social context. They may fail to interpret sarcasm, irony, culturally specific references, or historical nuance present in archival material. o Availability of Training Data: AI performance depends heavily on the availability of large, well-annotated training datasets. For highly specific or niche archival content, such datasets may not exist, limiting the applicability of pre-trained models and requiring costly annotation efforts. In summary, AI represents a transformative tool with enormous potential to improve the management, access, and exploitation of audiovisual archives. However, it is not a magic solution. Successful implementation requires a clear understanding of current capabilities and limitations. From a strategic perspective, this implies carefully selecting AI tools for specific tasks where expected benefits outweigh costs and risks, integrating these technologies with human expertise to ensure quality and relevance—particularly in quality control processes—and setting realistic expectations regarding performance, costs, and the need for ongoing evaluation. AI should be seen as a powerful complement within a broader archival strategy, not as a complete replacement for existing processes or expert human judgment.. Tabla 3: Aplicaciones de la IA en Archivos Audiovisuales AI Task Common Technology / Approach Potential Benefits for Archives Key Limitations / Challenges Transcription (ASR) Sequence-to-sequence models (Transformers, e.g. Whisper) Full-text search, Accessibility (subtitles), Indexing Variable accuracy (noise, accents, disfluency), Computational cost Object Recognition Convolutional Neural Networks (CNNs, e.g. ResNet, YOLO) Visual content-based search, Automatic tagging Accuracy depends on image quality and training data, Bias Facial Recognition Specialized CNNs Identification of individuals, Metadata linking Privacy and ethical concerns, Variable accuracy, Bias OCR in Video OCR models adapted for video On-screen text extraction (credits, headlines), Indexing Accuracy depends on resolution, font, motion Scene / Topic Segmentation Temporal models (ViViT), Image classifiers (ResNet), Multimodal analysis Content structuring, Improved navigation, Summaries Subjective definition of “scene”, Variable accuracy Sound Classification Audio models (AST, CNNs) Identification of music/speech/noise, Content filtering Accuracy in complex environments, Need for labeled data Speaker Identification (Diarization) Speaker embedding models (e.g. ECAPA-TDNN), ASR Attribution of dialogue, Participation analysis Accuracy in overlapping or noisy conversations Metadata Generation / Enrichment Combination of ASR, Visual/Audio Recognition, NLP (Language Models) Automated cataloguing, Linking to knowledge bases, Summaries Accuracy and relevance, Need for human validation, Contextual understanding   Chapter.4. Robust Metadata Strategies for Search and Management Metadata—data about data—are the cornerstone of effective management, discoverability, interoperability, and long-term preservation of digital audiovisual assets. Without adequate metadata, even perfectly preserved digital files risk being lost within vast digital repositories. .4.1. 4.1 Fundamental Audiovisual Metadata Standards: PBCore, EBUCore, MPEG-7 Given the complexity of audiovisual materials, specific metadata standards have been developed to capture their unique characteristics, going beyond generic schemas such as Dublin Core. • PBCore (Public Broadcasting Metadata Dictionary): o Origin and Focus: Created by the U.S. public broadcasting community, but applicable to audiovisual assets in general. It is based on Dublin Core, significantly extending it. o Structure:Organizes metadata into four main components: 1. pbcoreAsset / pbcoreDescriptionDocument: The top-level element representing the intellectual asset. 2. Intellectual Content: Descriptive metadata such as Title (pbcoreTitle), Subject (pbcoreSubject), Description (pbcoreDescription), Genre (pbcoreGenre). 3. Intellectual Property: Information about creators (pbcoreCreator), contributors (pbcoreContributor), publisher (pbcorePublisher), distributor (pbcoreDistributor), and rights (pbcoreRightsSummary). 4. Instantiation: Detailed technical metadata about each physical or digital manifestation of the asset (e.g. a specific tape, a digital file). Includes elements such as format (instantiationFormat), media type (instantiationMediaType), duration (instantiationDuration), file size (instantiationFileSize), data rate (instantiationDataRate), aspect ratio (instantiationAspectRatio), frame rate (instantiationFrameRate), audio/video codecs, and location (instantiationLocation – e.g. URL or physical identifier). An asset may have multiple instantiations. 5. Extension: Allows integration of metadata from other schemas (e.g. PREMIS, EAD). o Characteristics: Uses an XML schema (XSD) for validation and data exchange. Employs attributes to refine elements (e.g. titleType=”main”, startTime, endTime for temporal metadata). Well suited for detailed item-level description. • EBUCore o Origin and Focus: Developed by the European Broadcasting Union (EBU), also based on Dublin Core, but oriented toward broadcast workflows, content exchange, and semantic web applications. o Structure: Provides a framework for descriptive and technical metadata, often used in conjunction with the EBU Conceptual Class Data Model (CCDM). Published as EBU Tech 3293. o Characteristics: Includes extensions tailored to broadcast needs, such as the Audio Definition Model (ADM, adopted as ITU BS.2076) for immersive/object-based audio, advertising metadata (egtaMETA), camera acquisition metadata, and transport of identifiers such as ISRC in BWF files. Strong emphasis on semantics and interoperability through its RDF ontology (EBUCore RDF). An Iranian study confirmed the perceived importance of its elements among AV archive experts. • MPEG-7 (Multimedia Content Description Interface) o Origin and Focus: An ISO/IEC standard specifically designed to describe the characteristics of multimedia content (audio, video, images)—not the content itself—to enable efficient search, navigation, and filtering. Unlike MPEG-1/2/4, which encode content, MPEG-7 encodes descriptions. o Structure: Defines: 1. Descriptors (Ds): Representations of specific features (e.g. dominant color, texture, object shape, camera motion, voice pitch, musical rhythm, spoken keywords). 2. Description Schemes (DSs): Structures that specify relationships between Descriptors and other Description Schemes, enabling complex descriptions (e.g. temporal structure of a video, spatial relationships in an image). 3. Description Definition Language (DDL): An XML Schema–based language for defining new Descriptors and Description Schemes, ensuring extensibility. o Applications: Digital library and archive search, broadcast filtering, multimedia editing, content-based information retrieval (CBIR). It can represent ASR outputs and describe visual (color, texture, shape, motion) and audio (timbre, spoken content) characteristics. • Other Relevant Standards: o PREMIS: Fundamental for preservation metadata (see Section 1.4). o METS (Metadata Encoding and Transmission Standard): XML schema used to package descriptive, administrative, structural, and PREMIS metadata together with digital content files. o Dublin Core: Provides a basic set of 15 descriptive elements widely used as an interoperability baseline. o Specific Technical Schemas: Such as those defined by FADGI for embedded metadata in DPX or BWF, or schemas like AudioMD and VideoMD for detailed technical metadata. The choice of standard—or combination of standards—depends on the specific context and objectives of the archive. Dublin Core provides interoperable baseline description, while specialized standards such as PBCore, EBUCore, and MPEG-7 are essential for capturing the technical and content richness of audiovisual materials. PBCore excels at detailed asset and instantiation description; EBUCore is optimized for broadcast workflows and semantic interoperability; MPEG-7 focuses on deep content-feature description for advanced retrieval. In practice, a combination of standards (e.g. PBCore for description, PREMIS for preservation, MPEG-7 for content features) is often required. .4.2. Achieving Interoperability and Managing Rights Metadata The value of metadata increases exponentially when it can be shared and understood across systems and institutions (interoperability) and when it clearly communicates permissions and usage restrictions (rights metadata). • Interoperability Challenges: Heterogeneity is a major obstacle. Different archives may use different metadata standards, apply the same standards inconsistently, or rely on different controlled vocabularies. This complicates federated search, data exchange, and collection aggregation. Interoperability is not only technical (ability to exchange data) but also semantic (ensuring metadata meanings are interpreted consistently across contexts). • Strategies for Interoperability: o Adoption of Common Standards: Dublin Core, PBCore, EBUCore, PREMIS, METS. o Metadata Mapping (Crosswalking): Creating equivalence tables between schemas—complex and often involving information loss. o Conceptual Frameworks: Using high-level models such as OAIS to align understanding of functions and objects across systems. o Harvesting Protocols: Using OAI-PMH so aggregators (e.g. Europeana, DPLA) can collect metadata from multiple repositories. o Linked Data and Semantic Web: Using RDF, ontologies (e.g. EBUCore RDF), and persistent identifiers (URIs) to create machine-readable, interconnected, semantically rich metadata. • Rights Metadata: Rights metadata are critical but often poorly managed due to their complexity. They must clearly indicate: o Copyright Status: Public domain, copyrighted, orphan work. o Rights Holders: Who owns the rights (institution, third party, creator). o Permissions and Restrictions: Allowed and prohibited uses (viewing, copying, modification, commercial use, distribution), and under what conditions. This applies both to user access and to preservation actions (e.g. format migration). o Provenance Information: Date of rights research, responsible party. o Standards and Approaches:  Rights elements in general schemas (dc:rights, in Dublin Core, pbcoreRightsSummary in PBCore, PREMIS Rights entity).  Controlled Vocabularies: RightsStatements.org provides 12 standardized statements with URIs.  Machine-Readable Licenses: Creative Commons licenses and CC REL.  Blockchain and Smart Contracts: Emerging technology for recording ownership and automating permissions. A significant gap persists in rights metadata management. Despite its critical importance for preservation, access, and monetization, rights information is often insufficient, unclear, or difficult to interpret. Standardized, machine-readable approaches are essential to reduce legal and operational barriers and unlock reuse potential. .4.3. Implementing Effective Metadata Validation and Enrichment Workflows Metadata creation is not a one-time event; it requires well-defined workflows for initial validation and ongoing enrichment. • Metadata Validation: The process of ensuring metadata accuracy, consistency, completeness (per defined policies), and compliance with chosen standards. o Methods: Automated checks (schema validation, controlled vocabulary checks, date format verification) and manual review by cataloguers or subject experts. o Tools: Cataloguing systems, custom scripts, BWF MetaEdit, MDQC (Metadata Quality Control). o Importance: Validation ensures metadata reliability, directly improving search and archive management. • Metadata Enrichment: The process of adding value to existing metadata by expanding descriptions, adding technical detail, or establishing contextual links. o Sources: Additional cataloguer research, user contributions (collaborative tagging, comments), AI-based analysis (entity extraction, topics, summaries—see Section 3.2). o Examples: Adding identified personal names, linking locations to geographic coordinates, adding summaries, transcribing spoken content. • Integration into the Workflow: Metadata creation, validation, and enrichment must be integral to the digital object lifecycle: o Early Capture: Collect as much information as possible at creation or acquisition. o During Digitization: Automatically capture technical metadata; synchronize descriptive metadata creation. o Post-Ingest: Perform periodic validation, enable continuous enrichment, manage metadata updates. Metadata management must be treated as an ongoing lifecycle process aligned with the digital object itself. Metadata are not static; they require validation, enrichment, and migration over time. Neglecting active metadata management can render perfectly preserved digital objects undiscoverable or unusable, undermining preservation objectives. This continuous management is a core responsibility of the Data Management function within the OAIS framework. Digital preservation only reaches its full meaning when content can be located, accessed, and used effectively. Turning preserved archives into usable assets requires platforms, access models, and security criteria aligned with each organization’s cultural, operational, or economic objectives.  Chapter.5. Unlocking Value: Access, Distribution, and Monetization The digitization and preservation of audiovisual archives not only safeguards heritage, but also creates unprecedented opportunities for access, distribution, and value generation—whether cultural, educational, or commercial. This requires appropriate technology platforms, well-defined business models, and careful management of security and the user experience. .5.1. Enabling Access: MAM/DAM Platforms and Streaming Technologies For digitized archives to be useful, they must be accessible through systems that allow efficient management, search, and delivery. Digital Asset Management (DAM) and Media Asset Management (MAM) systems are the key technologies in this domain. • DAM (Digital Asset Management): o Scope: Broad-scope systems designed to manage a wide variety of digital assets, including images, documents, presentations, audio, and video. o Typical Users: Marketing, communication, sales, and human resources departments; organizations with diverse brand and content management needs. o Key Functionalities: Centralized storage, organization (categories, tags, metadata), search and retrieval, permission-based access control, version control, collaboration tools (review, approval), and distribution to multiple channels. o Potential Limitations for AV Content: They may struggle with very large video files or specific production/broadcast formats, and may lack advanced tools for video workflows (e.g. deep transcoding, tight integration with editing systems). They often focus on “hot” storage (fast access) and may not efficiently manage cold/archive storage. • MAM (Media Asset Management): o Scope: Systems specialized in managing rich media assets—primarily video and audio—optimized for production and post-production workflows. o Typical Users: Broadcasters, film and television production companies, post-production facilities, and audiovisual archives with large volumes of content. o Key Functionalities: Efficient handling of very large files (terabytes) and professional/broadcast formats; advanced metadata management (including time-based metadata); generation of low-resolution proxies for preview and remote editing; transcoding capabilities to multiple formats; integration with non-linear editing (NLE) systems; management of complex workflows; and often integrated hierarchical storage management (HSM), including cold storage tiers (e.g. LTO). o Advantages for AV Archives: Better suited to addressing the specific challenges of audiovisual content (file size, formats, complex metadata) and to supporting both preservation (long-term storage management) and reuse (production workflows). • Platform Selection: The decision between DAM, MAM, or an integrated solution depends on each organization’s strategic priorities. If the primary focus is internal management, long-term preservation, and production reuse, a MAM is likely more appropriate. If the main objective is broad distribution of diverse asset types (including marketing assets) to multiple users with strict brand control, a DAM may be considered—although it will likely require robust MAM capabilities for the audiovisual core. Aligning the platform with the archive’s objectives is crucial. • Streaming Technologies: Efficient streaming technologies are required for online delivery of audiovisual content. This involves: o Encoding for Access: Transcoding preservation masters into appropriate access formats (e.g. H.264/AAC in MP4) with bitrates optimized for web delivery. o Adaptive Bitrate Streaming (ABR): Technologies such as HLS or DASH that allow the player to dynamically select stream quality based on the user’s available bandwidth, ensuring smooth playback. o Delivery Infrastructure: Use of Content Delivery Networks (CDNs) to distribute content geographically and handle demand peaks. • User Access: Platforms must provide intuitive search and navigation interfaces adapted to different user profiles (researchers, creators, general public). Role-based permission management, aligned with rights metadata, is essential to control who can access which content and how it may be used. .5.2. 5.2. Diversifying Revenue: Licensing, FAST, AVOD, and ROI Considerations Digitization opens multiple avenues for generating revenue or recovering costs from audiovisual archives, going beyond traditional models. • Monetization Models: o Licensing: The traditional model in which specific usage rights are sold to third parties (e.g. broadcasters, production companies, advertising agencies, educational platforms) for inclusion in new works or distribution. This requires very clear rights management and structured negotiation and pricing processes. INA’s Ina Mediapro service is an example of large-scale professional licensing operations. o AVOD (Advertising Video-on-Demand): Offering content free of charge to end users, funded through advertising insertion (e.g. pre-roll, mid-roll, post-roll). Platforms such as YouTube, Pluto TV, Tubi, Peacock, and Roku Channel operate fully or partially under this model. It enables massive audience reach and monetization through programmatic and targeted advertising based on user data, and is particularly attractive to younger, streaming-native audiences. o FAST (Free Ad-Supported Television): Offering free thematic linear channels, programmed like traditional television but distributed via the internet (OTT), with revenue generated through advertising inserted during breaks. Platforms such as Pluto TV, Roku Channel, Samsung TV Plus, and Xumo are leaders in this space. It provides a familiar “lean-back” viewing experience for audiences accustomed to traditional TV and is ideal for large volumes of thematically grouped content (e.g. historical news, documentaries, classic series). o SVOD (Subscription Video-on-Demand): Access to a catalogue via a paid subscription (e.g. Netflix, Disney+). Less common for pure archives, but some services offer premium tiers or exclusive access to archival content as part of a broader SVOD offering. Hybrid models (lower-priced SVOD with advertising) also exist. o TVOD (Transactional Video-on-Demand): Pay-per-view or pay-per-download of specific content (rental or purchase). This may apply to highly exclusive or particularly high-value archival content. • Monetization Strategy: Selecting the appropriate model depends on factors such as content type, available rights, target audience, institutional objectives (maximizing revenue vs. maximizing access), and the organization’s capacity to invest in platforms and marketing. A deliberate strategy is essential; digitization alone does not guarantee revenue. A combination of models may be required (e.g. licensing premium content while offering curated selections via AVOD/FAST). The rapid growth of FAST and AVOD markets represents a particularly interesting opportunity for archives with large content volumes, providing scalable monetization paths beyond traditional licensing. • Return on Investment (ROI): Measuring ROI for digital preservation is complex but necessary to justify continued investment. ROI is not limited to direct monetization revenues. It also includes: o Cost Savings: Increased efficiency in internal workflows (search, retrieval), reduced physical storage costs, and prevention of losses due to degradation. o Brand and Reputation Value: Positioning as a trusted custodian of heritage and improved public image. o Cultural and Educational Impact: Enabling research, learning, and public engagement. o Enablement of New Services: The digital foundation allows the development of new offerings and products.The TCO analysis (Section 2.3) provides the cost baseline for calculating financial ROI. Table 4. Monetization Models for Audiovisual Archives Model Description Revenue Source Key Requirements Pros Cons Suitability for Archives Licensing Sale of specific usage rights to third parties. Direct licensing fees. Robust rights management, commercial/legal team, high-quality catalogue. High value per transaction, control over usage. Complex negotiations, limited reach, requires sales effort. High (for unique, historical, or high-quality content). AVOD Free on-demand content supported by advertising. Advertising revenue (programmatic/direct). Streaming platform, ad insertion technology (AdTech), content volume, audience. Broad potential reach, scalable model, attractive to users. Low revenue per view, dependence on audience size and advertising market, interrupted user experience. Medium to High (for general-interest content, documentaries, older series). FAST Free linear channels supported by advertising. Advertising revenue. Channel programming, thematic content volume, AdTech, distribution on FAST platforms. “Lean-back” viewing experience, passive discovery, continuous monetization. Less user control, requires large volumes to fill 24/7 channels. Medium to High (for large volumes of thematic content: news, sports, documentaries, classics). SVOD / Hybrid Subscription-based access (with or without advertising). Subscription fees (+ advertising revenue in hybrid models). Exclusive/compelling content, robust platform, subscription marketing. Predictable recurring revenue, higher perceived value. High barrier to entry for users, requires ongoing investment in content and marketing. Low to Medium (generally as part of a broader offering, not archive-only). TVOD Payment for specific content (rental/purchase). Transaction-based payments. Highly exclusive/high-demand content, e-commerce platform. High revenue per transaction for premium content. Less popular model for general archives, limited reach. Low (except for very specific, high-value material). .5.3. Ensuring Security, Optimizing the User Experience, and Exploring Federated Access The successful delivery of digital audiovisual archives depends not only on technology and business models, but also on ensuring asset security, providing an engaging user experience, and exploring collaborative access models. • Security: Security is a fundamental pillar for any digital archive, protecting assets against loss, corruption, or unauthorized access. Best practices include: o Physical Security: Protection of original media and local storage infrastructure (access control, environmental conditioning, protection against fire/flooding, perimeter security). o Digital Security:  Access Control: Robust user authentication and role- and permission-based authorization, ensuring that only authorized personnel can access or modify assets.  Encryption: Protection of data both at rest (in storage) and in transit (during transfer) using strong encryption algorithms.  Data Integrity: Use of checksums and regular integrity checks to detect any alteration or corruption of files.  Backup and Disaster Recovery (Backup & DR):  Implementation of robust backup strategies (e.g. the 3-2-1 rule), with copies stored in geographically separate locations and regularly tested recovery plans.  Network Security: Firewalls, intrusion detection/prevention systems, and security monitoring.  Vulnerability Management: Keeping software up to date and properly patched.  Standards Compliance: Adherence to recognized security frameworks such as ISO 27001, which defines an Information Security Management System (ISMS). Security must be an integral component of archive design and operation, as a breach can compromise decades of preservation effort and severely damage institutional reputation. • User Experience (UX): For access platforms—whether internal or public—a good user experience is crucial to encourage use and exploration of collections. This includes: o Intuitive Interface: Clear, easy-to-navigate design. o Effective Search and Discovery: Powerful search capabilities (full-text search, faceted search based on metadata), browsing options (by collections, themes, dates), and recommendations. o Reliable Playback: Stable video/audio players with good quality and standard controls. Consideration of ABR for streaming delivery. o Clear Metadata Presentation: Displaying relevant information about the content, its context, and its usage rights in a clear and understandable way. o Performance: Fast loading and response times. • Federated Access and Collaboration: Models that allow users to search and potentially access content from multiple distributed archives through a single interface, without requiring all content to be centralized. o Requirements: A high degree of metadata interoperability (see Section 4.2) and potentially standardized APIs (Application Programming Interfaces) for querying and retrieval. o Benefits: Dramatically expands the universe of content accessible to users, fosters collaboration between institutions, and avoids the costs and complexities of building a single massive repository. o Federated Learning: A related AI concept in which models are trained locally on decentralized data (within each participating archive), and only model parameters (not raw data) are shared to create an improved global model. This could be applied to collaboratively improve content analysis tools (e.g. ASR, object recognition) or recommendation systems in a federated environment, while preserving the privacy of each institution’s data. These federated approaches represent a promising path for large-scale inter-institutional collaboration, enabling knowledge sharing and improved discovery without the logistical and governance challenges of massive data centralization.  Chapter.6. Industry Implementations and Future Horizons The analysis of case studies from leading institutions and the exploration of emerging technologies provide valuable insight into current practices and future directions in the intelligent preservation and management of audiovisual archives. .6.1. Learning from the Leaders: Case Studies in Intelligent Archiving The field of digital preservation is constantly evolving, with new technologies and content types presenting both opportunities and challenges. • INA (Institut National de l’Audiovisuel, France): As the national archive of French broadcasting and the legal deposit institution, INA manages a massive collection (more than 18 million hours). Its strategy combines: o Large-Scale Digitization: An ongoing effort to digitize its analog heritage, driven both by preservation needs and by improved access. o Dual Access / Monetization Path: It provides free public access to a significant portion of the collection through its website (ina.fr) and licenses content to the professional market through Ina Mediapro. This demonstrates a balance between its public service mission and revenue generation (required to cover part of its budget). o Technological Infrastructure: Development of in-house systems for database management and consultation, accessible both in physical consultation centers and online. o Use of AI: Application of AI for content analysis and retrieval within its archives. o Collaboration: Agreements with organizations such as UNESCO to digitize and provide access to external collections. The INA case illustrates a mature model for managing a large-scale national archive, integrating preservation, research, public access, and commercialization. • BFI (British Film Institute) National Archive (United Kingdom): Custodian of one of the largest film and television collections in the world. o “Unlocking Film Heritage” Project: A key initiative to digitize 10,000 films from the BFI and regional archives, driven by the need to preserve analog heritage and make it digitally accessible. o Digital Preservation Infrastructure: They implemented a solution based on Spectra Logic T950 tape libraries and BlackPearl platforms. They adopted a “genetic diversity” strategy in tape media, using both LTO-6 and IBM TS1150 technology to mitigate risks associated with relying on a single vendor roadmap. o Integration and Access: The solution integrates with existing workflow applications and enables fast and easy access to digital files for global audiences. The BFI demonstrates a pragmatic approach to building large-scale digital preservation infrastructure, prioritizing long-term security, technological flexibility, and integration with workflows. • Library of Congress NAVCC (National Audio-Visual Conservation Center, USA): A cutting-edge facility built specifically to preserve the Library of Congress’s enormous audiovisual collections. o Digital File-Based Approach: A full transition to file-based digital preservation for sound and video, using high-capacity automated robotic systems for transfer. Film preservation remains primarily photochemical (cold storage), pending the maturation and affordability of film digitization technology. o Massive Infrastructure: Includes state-of-the-art preservation laboratories, cold storage for film (including separate vaults for nitrate), and enormous digital storage capacity (estimated at over 5 million gigabytes produced annually). The building includes sustainable features such as a large green roof. o Demand-Driven Access: Digital sound and video files are streamed to researchers in the Library’s reading rooms via fiber optics, meaning that user demand drives digitization and access priorities. o Center of Excellence: Conceived not only as a preservation center, but also as a hub for research, training, and international collaboration in audiovisual archiving. o The NAVCC represents a massive investment in infrastructure and technology, required to manage national-scale collections, and highlights the transition toward predominantly digital workflows for audio and video. • FIAT/IFTA “Save Your Archive” Program: An initiative by the International Federation of Television Archives to support smaller and at-risk audiovisual archives worldwide. o Approach: Provides funding and support for well-defined preservation, digitization, and valorization projects. o Selection Criteria: Prioritizes collections of high cultural value and high vulnerability risk, where the project has a significant impact on accessibility and is feasible in terms of budget, timeline, and team capacity. o Examples: It has supported diverse projects, from digitizing films about traditional crafts in Vojvodina or television archives in Madagascar, to preserving recordings of historic trials in South Africa or documentaries in Colombia. This program demonstrates the importance of international collaboration and support models tailored to archives with limited resources. • Other Notable Examples: o Google / EBU / CyBC Project: A collaboration in which Google provided state-of-the-art digitization facilities so that EBU members such as the Cyprus Broadcasting Corporation (CyBC) could digitize their tape archives free of charge (nearly 20,000 tapes for CyBC). This illustrates a public-private partnership model to address large-scale digitization. o Vanderbilt Television News Archive: An academic archive that applied ASR and AI tools to process and make its unique collection accessible. These case studies reveal a diversity of strategic approaches and operating scales. Large national institutions such as INA, BFI, and the Library of Congress make massive investments in tailored infrastructure and complex workflows, often balancing preservation with public access and, in some cases, commercial exploitation. On the other hand, collaborative initiatives such as FIAT/IFTA or EBU/Google offer alternative models to support smaller or less-resourced archives. There is no single correct model; the optimal strategy depends on collection scale, institutional mission, available resources, and the type of content managed. .6.2. Emerging Frontiers: Generative AI, Blockchain, Immersive Media Preservation, Federated Collaboration The field of digital preservation is constantly evolving, with new technologies and content types presenting both opportunities and challenges. • Generativa IA for Restoration: o Potential: Generative AI models can go beyond cleaning and correction by attempting to “recreate” lost details in low-resolution material (upscaling) or by adding color to black-and-white films. This can make historical content more appealing to contemporary audiences. o Challenges: The main concern is authenticity. AI does not recover lost details; it invents them based on learned patterns. This can introduce visual artifacts (the “AI look”) and create a representation that is not faithful to the original historical material. Its use for strictly archival purposes is questionable and requires extremely transparent documentation. Deepfake technology, which manipulates or synthesizes faces and voices, exemplifies the ethical and security risks associated with generative AI. • Blockchain for Trust and Rights: o Potential: Blockchain technology, due to its decentralized, immutable, and transparent nature, offers potential applications in digital archives for:  Integrity and Authenticity: Creating a tamper-proof record of files and their checksums, ensuring they have not been altered.  Provenance: Securely and verifiably tracking the history of a digital asset (creation, custody, modifications).  Digital Rights Management (DRM): Implementing smart contracts that automate and enforce usage licenses and access permissions transparently and without intermediaries. o Challenges: Scalability (handling large volumes of transactions), computational and energy cost, privacy (information on blockchain is often public or pseudo-anonymous), governance of decentralized networks, and the lack of mature legal frameworks and standards for its application in archives are significant obstacles. • Preservation of Interactive and Immersive Media (VR/AR): o Unique Challenges: Preserving Virtual Reality (VR) and Augmented Reality (AR) experiences is particularly complex due to their intrinsic dependence on a specific combination of hardware (headsets, sensors, controllers), software (graphics engines, operating systems, applications), user interfaces, and viewer interaction. The rapid cycles of technological obsolescence in this field are a major impediment. Preserving only the digital files is not sufficient; it is necessary to capture the “immersive essence” of the experience. o Emerging Strategies: Strategies are still under development, but include:  Comprehensive Documentation: Recording in detail the original hardware, software, configuration, and user experience.  Preservation of Physical Artifacts: Retaining the original hardware whenever possible.  Emulation: Creating software that simulates the original hardware and software environment to run the application (as is done with old video games).  Migration / Recreation: Adapting or recreating the experience using contemporary platforms and technologies.  Video Capture: Recording the experience from the user’s perspective (although this loses interactivity and immersion). Collaboration between creators, technologists, and institutions is vital. General digital preservation principles also apply, such as file format migration and metadata management. • Federated Collaboration and Federated Learning: o Potential: Enables multiple institutions to collaborate on computationally intensive tasks or those requiring large datasets, without needing to centralize raw data. o Federated Learning: Specifically, it enables collaborative training of AI models (e.g. to improve ASR, image recognition, etc.). Each institution trains the model using its own local data (preserving privacy and control), and only model updates (parameters, gradients) are shared and aggregated to create a more robust and generalizable global model. o Applications: It is already used in fields such as remote sensing and healthcare. It could be applied to AV archives to collaboratively improve content analysis tools or recommendation systems, overcoming the limitations of data from a single institution. These emerging frontiers indicate that the field of audiovisual preservation must be adaptable and innovative. Technologies such as generative AI and blockchain offer promising solutions to persistent problems (restoration, trust), but also introduce new technical complexities and ethical dilemmas (authenticity, privacy). The emergence of new formats such as VR/AR requires the development of entirely new preservation methodologies. Federated approaches suggest a future of increased distributed collaboration. In this context, it is essential for organizations to selectively evaluate the applicability of these trends and to work with technology partners capable of supporting the definition of future-ready preservation strategies, such as Telefónica Servicios Audiovisuales (TSA).  Chapter.7. Strategic Recommendations for Audiovisual Digitization and Preservation Projects Digitizing and preserving audiovisual archives requires a structured, long-term approach. Experience gained from projects of different scales and nature shows that sustainable results do not depend solely on the technology used, but on the consistency between strategy, processes, standards, and organizational capabilities. Based on the analysis developed throughout this document, a series of key recommendations can be identified for organizations that are considering or initiating digitization, preservation, and advanced management initiatives for audiovisual archives. .7.1. Conduct a Comprehensive Risk Assessment and Prioritization Any digitization initiative should begin with a systematic assessment of the risk associated with existing collections. This assessment must jointly consider the physical degradation of the media, the technological obsolescence of formats and equipment, and the value of the content from heritage, operational, or commercial perspectives. Risk- and value-based prioritization makes it possible to define realistic, phased action plans and allocate resources efficiently, avoiding reactive or arbitrary approaches. .7.2. Standardize Workflows and Preservation Formats Early standardization of digitization workflows is a critical factor for quality, operational efficiency, and long-term sustainability. Defining and documenting processes aligned with international best practices reduces variability, facilitates quality control, and simplifies future infrastructure evolution. Likewise, selecting open, widely adopted, and well-documented preservation formats helps minimize risks of obsolescence, technological dependency, and interoperability difficulties.   .7.3. Design a Sustainable and Secure Storage Architecture Storage-related decisions must be based on a long-term vision and a rigorous analysis of total cost of ownership. In most contexts, hybrid architectures combining different storage tiers are particularly suitable for balancing access, security, resilience, and cost requirements. It is essential to define clear policies for replication, periodic data integrity verification, media refresh, and technological migration planning as an integral part of the preservation strategy, rather than as isolated actions. .7.4. Rely on Appropriate Audiovisual Asset Management Platforms Efficient management of digital audiovisual archives requires platforms capable of handling large volumes of content, complex metadata, and workflows specific to the audiovisual domain. The selection of these platforms should consider their ability to integrate with hierarchical storage infrastructures, their compatibility with metadata standards, and their flexibility to incorporate new functionalities over time. These capabilities can be implemented either through in-house infrastructures or through managed services, provided they incorporate deep knowledge of the processes involved in digitization, preservation, and reuse of audiovisual content. .7.5. Explore the Use of Artificial Intelligence Progressively and in a Controlled Manner Artificial intelligence offers relevant opportunities to improve efficiency in the management of audiovisual archives, particularly in metadata generation and enrichment. However, its adoption should be approached gradually, prioritizing specific use cases and carefully evaluating its impact in terms of accuracy, cost, and governance. Integration of these technologies is most effective when conceived as support for human teams, incorporating validation mechanisms that ensure the quality, context, and traceability of the generated information. .7.6. Define a Comprehensive Metadata and Rights Management Strategy Metadata are a cross-cutting element that influences preservation, interoperability, access, and content reuse. Adopting established standards and defining clear application profiles makes it possible to describe and manage audiovisual assets in a consistent and sustainable manner. Likewise, it is essential to establish a clear framework for managing the rights associated with content, facilitating both long-term preservation actions and potential access and reuse models, while reducing legal and operational risks. .7.7. Integrate Access and Reuse Considerations from the Outset Digital preservation reaches its full meaning when content can be located, consulted, and used effectively. Integrating considerations related to access, user experience, and potential reuse models from early stages helps align technical decisions with the cultural, educational, or economic objectives of each initiative. This approach helps maximize the impact of the digital archive and avoid investments that, although technically sound, may limit the future use of the content. .7.8. Promote Continuous Training and Ongoing Technology Monitoring Audiovisual preservation is a dynamic field in which standards, tools, and formats evolve continuously. Investing in ongoing training for the teams involved and maintaining active monitoring of technological developments in the sector are key factors to ensure the resilience and adaptability of implemented solutions. The combination of specialized knowledge, continuous updating, and a strategic approach makes it possible to address audiovisual preservation as a sustainable and future-ready program. An approach based on continuous training and technological updating helps ensure the resilience and adaptability of the implemented solutions.  Chapter.8. Conclussion The digitization and preservation of audiovisual archives now constitute an unavoidable challenge for many public and private organizations. The physical degradation of media, technological obsolescence, and the progressive loss of playback capabilities place a significant portion of audiovisual heritage in a situation of real and irreversible risk. In this context, inaction implies accepting increasing losses of cultural, operational, and economic value. Throughout this White Paper, it has been shown that effectively addressing audiovisual preservation requires moving beyond isolated project-based approaches and adopting a vision of a continuous, structured, and actively managed program. This approach is built on several fundamental pillars: the adoption of established reference frameworks such as OAIS; the use of open metadata and preservation standards; the implementation of rigorous and controlled digitization workflows; the design of sustainable storage architectures based on total cost of ownership criteria; and the progressive integration of technologies such as artificial intelligence to improve efficiency, access, and content reuse. Likewise, the document has emphasized that digital preservation only reaches its full meaning when it is linked to clear strategies for access, rights management, and content valorization. The ability to locate, understand, and reuse audiovisual archives transforms preservation from a technical obligation into a strategic lever that strengthens institutional relevance, operational efficiency, and the creation of new cultural and economic opportunities. In a technological and regulatory environment that is constantly evolving, audiovisual preservation is becoming established as a strategic decision that requires a long-term vision and the coherent integration of technologies, processes, and standards. Successfully addressing this challenge involves combining technical capabilities, knowledge of the audiovisual ecosystem, and management capable of sustaining the preservation program over time. Within this framework, organizations with experience in complex audiovisual digitization and preservation projects play a key role as specialized technology partners and integrators. Telefónica Servicios Audiovisuales (TSA) is positioned in this space, contributing expertise in the definition and implementation of preservation strategies, as well as in the integration of infrastructures, platforms, and services tailored to the specific needs of each organization.  Chapter.9. Glossary of Key Terms • AIP (Archival Information Package): Archival Information Package. A set of information (content object and preservation metadata) that is stored and preserved within an OAIS. • ASR (Automatic Speech Recognition): Artificial intelligence technology that converts spoken audio into text. • AVOD (Advertising Video-on-Demand): Advertising Video-on-Demand. A business model in which users access video content free of charge in exchange for viewing advertisements. • BWF (Broadcast Wave Format): An extension of the WAV file format that allows additional metadata to be embedded; commonly used for audio preservation masters. • Checksum: A fixed-size value calculated from a block of digital data to detect errors or modifications (used in fixity checks). • DAM (Digital Asset Management): A system for storing, organizing, managing, and distributing a wide variety of digital assets. • DIP (Dissemination Information Package): Dissemination Information Package. An information package derived from one or more AIPs, prepared by the OAIS for delivery to a consumer. • DPX (Digital Picture Exchange): A file format commonly used to store individual frames of scanned film at high quality. • EBUCore: A metadata standard developed by the European Broadcasting Union to describe audiovisual resources, particularly in broadcast contexts. • FAST (Free Ad-Supported Television): Free Ad-Supported Television. A content distribution model that offers scheduled linear channels over the internet, funded by advertising. • FADGI (Federal Agencies Digitization Guidelines Initiative): A U.S. federal agencies initiative that develops guidelines and best practices for the digitization of cultural heritage materials. • FFV1: A lossless video compression codec, often used within the MKV container for video preservation. • Fixity: A property of a digital object indicating that it has not been altered or corrupted. Verified using checksums. • AI (Artificial Intelligence): A field of computer science dedicated to creating systems capable of performing tasks that normally require human intelligence, such as pattern recognition, learning, and decision-making. • Generative AI: A subfield of AI focused on creating new content (images, text, audio) based on the data on which the models have been trained. • IASA (International Association of Sound and Audiovisual Archives): An international association that publishes technical guidelines for the preservation of audio and video. • Ingest: An OAIS function responsible for receiving Submission Information Packages (SIPs) from producers and preparing them for archival storage as AIPs. • ISO 27001: An international standard for Information Security Management Systems (ISMS), relevant to the security of digital archives. • LPCM (Linear Pulse Code Modulation): A standard method of encoding uncompressed digital audio, commonly used in WAV and BWF files. • LTO (Linear Tape-Open): An open-standard magnetic tape technology widely used for backup and long-term data archiving. • MAM (Media Asset Management): A system specialized in managing video and audio assets, often including functionalities for production and post-production workflows. • METS (Metadata Encoding and Transmission Standard): An XML schema for encoding descriptive, administrative, and structural metadata about objects within a digital library, used to package digital objects. • MKV (Matroska): An open-source, flexible multimedia container format capable of storing multiple video, audio, and subtitle tracks. Often used with FFV1 for preservation. • MPEG-7: An ISO/IEC standard for describing multimedia content characteristics to facilitate search and retrieval. • MXF (Material Exchange Format): A professional audiovisual container format commonly used in broadcast and post-production environments. Often used with JPEG 2000 for archival purposes. • OAIS (Open Archival Information System): A reference model (ISO 14721) that defines a conceptual framework for an archival system dedicated to the long-term preservation and access of digital information. • Technological Obsolescence: The process by which technology (hardware, software, formats) becomes outdated or unusable, making access to stored information difficult. • OCR (Optical Character Recognition): A technology used to convert images of text (scanned or within video) into editable and searchable text. • PBCore (Public Broadcasting Metadata Dictionary): A metadata standard based on Dublin Core, designed to describe audiovisual assets, particularly in the public broadcasting context. • PREMIS (PREservation Metadata: Implementation Strategies): A data dictionary (ISO 22957) that defines the core metadata required for the long-term preservation of digital objects. • QC (Quality Control): Processes used to verify that digital files meet required technical and content specifications. • ROI (Return on Investment): A metric used to evaluate the efficiency or profitability of an investment. • Sticky Shed Syndrome: Degradation of the binder in magnetic tapes due to hydrolysis, making them sticky and difficult to play back. • Vinegar Syndrome: A chemical degradation process affecting cellulose acetate film, releasing acetic acid (vinegar smell) and causing brittleness and shrinkage. • SIP (Submission Information Package): Submission Information Package. An information package delivered by the producer to the OAIS for ingest. • TCO (Total Cost of Ownership): A financial estimate designed to help evaluate the direct and indirect costs related to the acquisition and operation of an asset (such as a storage system) over its lifecycle.   The conception and strategic direction of this TSA white paper are the result of the knowledge and experience of the Telefónica Servicios Audiovisuales team. During the development process, we used various artificial intelligence tools—such as Microsoft Copilot for research and initial information exploration, as well as Gemini Deep Research, ChatGPT 5.2, and Claude to optimize review and analysis stages. The final evaluation, aimed at ensuring relevance for our audience, relied on the capabilities of other specialized artificial intelligence solutions to emulate audience characteristics. Nevertheless, the core ideas, in-depth analysis, and conclusions presented in this document are the result of the expertise and professional judgment of our team.