Moving Picture, Audio and Data Coding by Artificial Intelligence – MPAI – has been devised as an international non-profit organisation with the mission to take over the baton from old-style compression to the new AI-based compression . This will take compression performance to new levels, extend the benefits of compression to all industries beset by huge amounts of data and give them the possibility not only to save costs from compression, but to get more out of their data.
Now that MPAI has been officially constituted on Wednesday 30 September 2020 (see Press Release), what will MPAI do?
This is a reasonable question to ask, but a better question would be: what has MPAI been doing? This is because, some 2 months before its actual establishment, a group of highly motivated experts has developed some use cases, aggregated in areas where MPAI standards can make the difference.
Thanks to the efforts of many, MPAI has the road already mapped out with several activities at different levels of maturity. The list below gives the more mature areas of the many that have been explored (see the list of use cases). The list order is a personal assessment of the maturity.
- Context-based Audio Enhancement (MPAI-CAE) is the most mature area. By using AI, MPAI-CAE can improve the user experience in a variety of instances such as entertainment, communication, teleconferencing, gaming, post-production, restorarton etc. in a variety of contexts such as in the home, in the car, on-the-go, in the studio etc.
- Integrative AI-based analysis of multi-source genomic/sensor experiments aims to define a framework where free and commercial AI-based processing components made available in a horizontal market can be combined to make application-specific “processing apps”.
- Multi-modal conversation aims to define an AI-based framework of processing components such as fusion of multimodal input, natural language understanding and generation, speech recognition and synthesis, emotion recognition, intention understanding, gesture recognition and knowledge fusion.
- Compression and understanding of financial data aims to enable AI-based filtering and extraction of key information from the flow of data that companies receive from the outside, generate inside or issue because of regulatory compliance.
- Server-based Predictive Distributed Multiplayer Online Gaming aims to minimise the visual discontinuities experienced by gameplayer by feeding the data collected from the clients involved in a particular game to an AI-based system that can predict each individual participants’ moves in case that information is missing.
- AI-Enhanced Traditional Video Coding aims to develop be a video compression standard that will substantially enhance the performance of an existing video codec by ehnancing or replacing traditional tools with AI-based tools.
MPAI signals a discontinuity with the past not only in the technology it uses to address known industry needs, but also in the way it overcomes the limitations of the Fair, Reasonable and Non-Discriminatory (FRAND) licensing declarations, a burning issue for many standard developing organisations and their industries. MPAI plans to develop and make known, for each MPAI standard, a “framework licence”, i.e. the business model, without values, dates and percentages, that standard essential patent holders intend to use to monetise their patents adopted in the standard.
Companies, academic institutions and individuals representing departments of academic institutions may apply for MPAI membership, provided that they can contribute to the development of technical specifications for the efficient use of data.