The climate crisis is at hand. Our generation will likely know the success or failure of our efforts to address human impacts on the environment. The next generation will look back on what we have done: a future that we can make but not yet fully know; a past they will know but not any more change. This keynote will discuss the prospects for digital remembering of a time before the crisis for the generations that come after it. Their perspectives will depend on the records and the archives which we create, curate and hand on to them.  Standing in the way are the numerous challenges which we face in developing economically and environmentally sustainable preservation of digital materials. The presentation will end with a call to action, to secure the inheritance of our digital collections.  The preservation community has never not cared for the generations that will follow, but there’s never been more important time to take our stand with the future.

Transforming and researching the news: the National Corpus of Irish project – The National Corpus of Ireland project is an initiative to develop a large national corpus of contemporary Irish encompassing both written and spoken sources. Various subcorpora and related specialised corpora will also be created in what will become a hub for corpus-based research on the Irish language. In this talk, the need for such a resource will be explained, the steps in the project will be set out, and we will look at some examples of research that might result from it. We will focus on the capture and reuse of news archives (RTÉ, Irish Times, Tuairisc, TG4 and others) and the possibilities afforded by audiovisual archives for corpus development. Compilation of this corpus is being undertaken by the Gaois research group, Fiontar & Scoil na Gaeilge, DCU, with funding for the period 2022-2024 from the Department of Tourism, Culture, Arts, Gaeltacht, Sport and Media.

The carbon footprint of the ICT sector has been stable around 0.7 billion ton CO₂e for nearly a decade. Even as more people than ever use a smartphone or mobile phone (8+ billion), and the Internet (5+ billion). Energy efficiency improvements has countered the growth in number of users. Before the Covid-19 pandemic, the aviation sector’s carbon footprint was about 3 times larger, 2.2 billion ton including so called high-altitude effects, for serving only about 10% of all humans a few times per year. That the ICT sector’s carbon footprint is larger than aviation is based on an old erroneous comparison.

During the pandemic, the ICT sector helped the world with working online from home, online education, online shopping for, e.g., food, and with support to the health care system to handle the pandemic. And then with all forms of online entertainment and media during the lockdowns. Total data traffic increased by about 50%, yet the footprint stayed nearly the same.

It has, however, also been claimed in media that streaming video, or downloading digital media, or having videoconferences, consumes a lot of electricity. These claims are not supported by any real measurements. It would have been easy to measure e.g., a Wi-Fi router and a laptop during a video conference or when streaming video and seeing a small power increase. Even downloading a large PC game consumes only a few Watts over Wi-Fi depending on signal quality and significantly less using a network cable.

Netflix and its use of AWS CDN (Content Delivery Network), together with Akamai, represents about 30% of total Internet data traffic. Yet their data centers and CDNs consume only about 1.1 TWh. Meta/Facebook consume over 9 TWh for about 15% of total Internet data traffic, showing the large difference between only storing and sending data, compared to also processing large amounts of data.

This talk explores the challenges and opportunities surrounding the design of sustainable infrastructures for the storage of large digital collections. Rather than approaching sustainability from a traditional environmental standpoint, this paper uses the lens of sustainable infrastructures to explore the value of long-lasting, modular, repairable, and extensible digital storage. Using the experience of designing an archival storage system for the ITMA as a case study, the aim is to show that sustainable infrastructures may be a promising approach used to contribute to the goals of sustainability in archives.

Meemoo manages a large quantity of mainly audiovisual material from more than 170 partners in cultural heritage and media. More than 6 million objects are currently stored, ranging from digitised newspapers, photos, videos, and audio. In addition, a number of access platforms make the digitised content available to specific target groups, including teachers, students, professional re-users, or the public.

Metadata is a key element in all of meemoo’s processes. An important part of our activities is to collect, integrate, manage, and search a large variety of heterogeneous metadata across the archived content. The scale of this has increased enormously, so a good and integrated approach is needed to deal with the amount of metadata, its need for flexibility, and how easy it is to find. One of the specific challenges is modelling and storing data from machine learning algorithms (speech recognition, face detection and entity recognition) for reuse.

In this talk, we will discuss the key points and lessons learned from implementing the new metadata roadmap at Meemoo, which is focused on a Knowledge Graph-based infrastructure. The goal of the roadmap is to establish a better data practice within the organization and offer application-independent, uniform access to (meta)data that is spread across various systems and formats.

In this talk, we would like to discuss questions like:

• How many files do we store and why?
• What should we know about file quality?
• Which factors affect file quality, and what has the archive got to do with it?
• Short-, Mid- and Long-Term Strategies to deal with imperfect files?

The talk will not go into technical details but will focus on the aspects concerning archive media management.

Sound & Vision’s archiving activities traditionally revolve around ingesting and preserving the radio and television of Dutch public broadcasters. However, as a result of different fusions, acquisitions and tenancies, different types of materials (both audiovisual and non-av) and accompanying non-standardized metadata are becoming a steadily larger chunk of our collections. Sound & Vision will present the impact of these changes in terms of metadata ingest, thesaurus use and overall work practices.

In this talk, we will explore how we changed persons and organisations from uncontrolled metadata to a controlled vocabulary during the migration of metadata from an old MAM system to a new in-house built MAM.

RTÉ Archives is in a process of digital transformation. This presentation will look at our approach and progress with digitisation, digitalisation and digital transformation; how we are restructuring our teams and developing new skills and juggling both digital and analogue workflows in parallel; and what opportunities and challenges will arise with the provision of digital access to our archive content in the future.

It will also look at RTÉ’s new digital storage solution examining the options considered and the approach taken to consolidate disparate digital storage systems into one system that is sustainable, resilient, secure and scalable. It will explain how object storage manages resilience and security and how it facilitates RTÉ Archives’ storage and digital preservation requirements.

Content preservation, high-quality production and process automation are at the core of the current transformation of Public Service Media (PSM) from its traditional business to the modern digital era. Broadcasters’ archives play a central role in this change, being a valuable source of information to run the business. On the other hand, accessing and retrieving desired content (and metadata) may be a non-trivial task. This is where artificial intelligence comes into play, providing solutions that help users extract knowledge and organise data more efficiently and effectively.

This presentation highlights how emerging AI-based technologies can support the mission of PSMs in this transition by exploiting AI’s potential to facilitate archive organisation and valorisation. Sample tasks in this area are those aimed at AI-based metadata extraction, like Natural Language Processing, geographical landmark recognition or TV celebrity identification.

In this context, we describe a use case based on the main tasks usually performed by media professionals during their everyday job. To validate our vision, we collected opinions from industry experts, which let us rank AI functionalities according to users’ needs for their implementation.

AI technologies have been used at the RTS for automatic metadata extraction for several years. Currently, we are in the final stages of implementing speaker recognition. This presentation will show how documentalists have been involved and the role they play in the development and use of these new tools.

The RTÉ archives house vast amounts of analogue audio/-visual material from 20th-century radio and television broadcasting, which are being digitised as part of an ongoing initiative.

These materials are associated with varying degrees of metadata, but in the absence of transcription, the raw data (i.e. audio) cannot be searched. While human transcription remains the gold standard in terms of accuracy, automatic speech recognition (ASR) systems offer a speed and a scale that can be used in tandem with transcribers to improve workflow efficiency.

ASR has been thoroughly researched and developed for major languages like English, but the Irish-language equivalent is still at an early stage. Although impressive results have been achieved for careful read speech, performance is markedly worse for conversational, spontaneous speech, the dominant style found in broadcast materials. This is primarily due to a lack of sufficiently large Irish speech corpora. However, recent advances in self-supervised learning have seen the inclusion of unlabelled data in training corpora i.e. audio materials without transcriptions, significantly reducing the amount of paired audio-transcription data required to achieve good performance. Thus, the application of research-stage Irish ASR systems to unlabelled speech corpora in, e.g., the RTÉ Archives presents an opportunity to accelerate both (i) the training of these systems and (ii) the conversion of these data to a format more accessible for cultural and academic purposes. This, in turn, can contribute to ongoing linguistic research, the results of which may themselves be fed back into speech technology development.

In many types of content, lower-third graphics contain valuable information. Up until 2022, archivists at VRT used lower thirds character graphics as a visual source of data for manual annotation. This meant scrubbing through footage, looking for occurrences of lower-third graphics, and entering the metadata manually into the system. In this presentation, Marijn Daniels will explain how VRT used A.I. to automate the process of speaker labelling, using lower-third character recognition.

For the last decade INA has engaged in a wide digital transformation. The fresh redesign of our data model quickly allowed to experiments with AI based tools such as machine learning. This speech aim to present the solution we built to assist the documentary processing of TV news channel and will focus on the business aspects. What changes do recent technical progresses bring for the librarians and heritage community?

Following an intensive digitisation process of their archives, M6 launched 2015 a B2B portal to expose their legacy production videos to potential business customers. Within five years, all the key objectives set in the original plan were reached, and the archive unit turned from a cost centre into a profit centre. What lessons were learnt from 7 years of operating such a content sales portal? What organisation was put in place? Today M6 is gearing up their approach by launching a new portal allowing it to expose and sell all media produced by the group (audio, pictures, etc.). The aim is to double the revenue and activate a new monetisation stream of their legacy assets. The session will cover the initial project M6VideoBank in 2016, our seven years of experience, and the improvement of the portal and its new version with the Perfect Memory technology.

The covid crisis motivated ERT Archives to transform itself in the past two years not only by adjusting to the new distance-working conditions but also by updating and adopting a MAM that quickly became the basic tool for the ERT S.A. workflow operations (Dalet). Within two years, ERT S.A. as a whole transformed into a fully digital working environment, and specifically ERT Archives are undergoing a considerable transformation with a digitization program of more than 200.000 beta tapes that is about to begin. The challenges are many, and the lessons to be learned are equally important: How is the sustainability of an Archive maintained while its primary goal is to facilitate the broadcaster’s day-to-day needs? How can a broadcasting MAM be optimized and customized to serve as an archive’s MAM? What is the cost of speed of operations? How are Archival requirements prioritized? How can planning solve digital storage issues, and how can training ensure a minimum of Metadata quality standards? The presentation will explain the 2-year timeline of this transformation and its technical aspects, milestones and results, prospects and lessons. It will also describe ERT Archives’ new digitization program and discuss the challenges and issues that need to be addressed in order to optimize operations and maintain the Archives’ standards, along with the new roles that ERT’s Archivists are called upon to assume.

At Sound & Vision in 2012, speaker labelling and, later in 2019, face recognition was implemented and hooked up to the broadcast archive in an effort to reduce time spent on manual annotation. Also, in the training of models and correction of the AI output, manual intervention by archivists is minimal.

At RTVE, the integration of massive automatically generated data only started in 2021 after some years of training. Since then, archivists have been involved in a complex process to discern the optimal application of these AI tools in our daily work and understand the overall impact on the archive.

In this presentation, RTVE & Sound & Vision will compare their use of AI and human annotation practices.

A reflection on the large-scale audio content digitisation (and migration) from magnetic tape and optical disk carriers with an emphasis on knowledge and skills -nascent and prevailing- throughout the project’s key activities of planning, pre-digitisation, digitisation, and post-digitisation.