We’re obsessed with challenging ourselves to come up with better, faster, stronger – and more accurate – solutions for what we do. That’s why we invest over 3 million euros every year in our in-house Research and Development.
Our Proprietary Audio Technologies
We match the fingerprint of any audio against the 500 years worth of indexed music in our database. Our fingerprints maintain accurate performance in conditions of all types – interruptions, speech, sound effects, for example – by running on the selection of discriminative robust events. It is also used for copyrighting verification.
We match the metadata from digital sales providers with the internal music databases of collecting societies and enrich it with our own database to ensure it is complete. Every day we deliver 27 billion matches to make sure music creators receive the royalties they deserve.
Using deep learning tech, we scan audio recordings to detect cues that show the presence of music. For every 2-second segment, we output a value to represent both the musical and the simultaneous non-musical sounds present.
The audibility estimation determines whether the background music in a TV programme, show or advert is “audible” or “inaudible”. The specifications are given by the client in each case.
We use our well-trained algorithm to confirm whether any 2 audio files are the same musical work – when processed on a song vs song basis – or not.
An audio recording is processed and evaluated. A sequence of identified notes corresponding to the audio recording is determined by iteratively identifying potential notes within the audio recording. A rating for the audio recording is determined using a tuning rating and an expression rating. The audio recording includes a recording of at least a portion of a musical composition.
Methods and devices for estimating audibility of audio samples in audio mixes of broadcast media programs are proposed. Example methods comprise the steps of providing a representation matrix of an audio mix, the audio mix comprising the audio sample and additional audio, providing a representation matrix of the audio sample, subtracting the representation matrix of the audio sample from representation matrix of the audio mix to generate a difference matrix, applying an audibility model to the difference matrix to generate an audibility matrix, determining an audibility level for each element of the audibility matrix, and averaging the determined audibility levels of the matrix to estimate the audibility of the audio sample.
Our Research And Innovation Projects
FuturePulse – Multimodal Predictive Analytics and Recommendation Services for the Music Industry – has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 761634
Bloomen – Blockchains in the new era of participatory media experience has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 762091.
The National Research National Agency of the Ministry of Science and Innovation (La Agencia Estatal de Investigación del Ministerio de Ciencia e Innovación) supports the Retos-Colaboración 2019 project New generation of music monitoring technology (NextCore), RTC2019-007248-7.
Identi@rt – R+D en preservació del valor d’actius digitals de música, art I disseny, amb distributed ledger tech – (COMRDI18-1-0011) of la Comunitat Media (COM18-1-0002) is financed by ACCIÓ (Agència per la Competitivitat de l’Empresa, Generalitat de Catalunya) as part of the collaborative projects for investigation, development and innovation of the Communities RIS3CAT.
ACCIÓ (Agència per la Competitivitat de l’Empresa, Generalitat de Catalunya) supports the project LoudSense – AI system for automatic audibility estimation of background music in audiovisual productions (ACE014/20/000051) as part of the call INNOTEC 2020.
PICAE – Publicació Intel·ligent de Continguts Audiovisuals I Editorials – (COMRDI18-1-0007) of la Comunitat Media (COM18-1-0002) is financed by ACCIÓ (Agència per la Competitivitat de l’Empresa, Generalitat de Catalunya) as part of the collaborative projects for investigation, development and innovation of the Communities RIS3CAT.
Ministerio de Economía y Empresa (Ministry of Economy and Enterprise) supports the Strategic Action of Economy and Digital Society – the Digital Enabling Technologies impulse 2019, project System for real-time massive audio monitoring based on artificial intelligence (DeepTrack), TSI-100903-2019-38.
Ministry of Industry, Energy and Digital Agenda (Ministerio de Energía, Turismo y Agenda Digital) supports the Strategic Action of Economy and Digital Society – Technological Impulse 2017, Project PlayIT: Big Data BI scalable engine for the process and identification of music metadata – TSI-100600-2017-20). Co-funded by Fondo Europeo de Desarrollo Regional (FEDER) – The way to build Europe.
The projects Music/Speech detection in broadcast media programs (DI46-2016) and Music identification algorithms using deep learning techniques (DI46-2020) are supported by the Industrial Doctorates plan of the Secretary of Universities and Investigation of the Department of Companies and knowledge of the Catalonia Generalitat (Secretaría de Universidades e Investigación del Departamento de Empresa y Conocimiento de la Generalitat de Cataluña).
Red.es (Ministry of Economic Affairs and Digital Transformation) supports the Strategic Action of Economy and Digital Society – the Digital Enabling Technologies impulse 2020, project Cloud Computing Platform for Scalable and Elastic Musical Monitoring (BigWave), 2020/0720/00097949. Co-funded by the European Union through the European Regional Development Fund (FEDER) – The way to build Europe
In 2018, MIREX -the most important international competition of MIR (Music Information Retrieval) algorithms- recognised our music detection algorithm as the best in the field, with accuracy 10 percentage points higher than the algorithm in second place. In 2019, our first position was reestablished as we presented 2 new algorithms and they both obtained better results in comparison to the algorithm submitted in 2018.
BMAT was appointed as ‘Key Innovator’ by the European Commission’s Innovation Radar for the developments in the H2020 Bloomen project: Blockchain technology will be used to provide copyright protection of media content.
Kapsoulis, Nikolaos; Psychas, Alexandros; Palaiokrassas, Georgios; Marinakis, Achilleas; Litke, Antonios; Varvarigou, Theodora; Bouchlis, Charalampos; Raouzaiou, Amaryllis; Calvo, Gonçal; Escudero Subirana, Ordi. 2020. “Consortium Blockhain Smart Contracts for Musical Rights Governance in a Collective Management Organizations (CMOs) Use Case“
Future internet 12, no. 8: 134.
Meléndez-Catalán, B. Molina, E., and Gómez, E. (2017). “BAT: an open-source web-based audio-events annotation tool“.
3rd Web Audio Conference
Meléndez-Catalán, B. Molina, E., and Gómez, E. (2019b). “Open broadcast media audio from TV: A dataset of TV broadcast audio with relative music loudness annotations“.
Transactions of the International Society for Music Information Retrieval
Meléndex-Catalán, B. Molina, E., and Gómez, E. (2020). “Relative music loudness estimation, using temporal convolutional networks and a CNN feature extraction front-end“.
In Proceedings of the 23rd International Conference on Digital Audio Effects (DAFx-20), volume 5, pages 273-280.
Our Scientific Publicactions and Awards