Project Details

Exchanges for SPEech ReseArch aNd TechnOlogies

Project Period: 1. 1. 2021 - 31. 12. 2025

Project Type: grant

Code: 101007666

Agency: European Comission EU

Program: Horizon 2020

Czech title
Výměny pro výzkum řeči a technologií
Type
grant
Keywords

artificial intelligence, intelligent systems, multi agent systems, machine learning, data mining, statistical data processing and application, modelling engineering, human computer interaction, natural language processing, speech processing, neural networks, explainability, human assisted learning, low resources, natural language processing, standardization, evaluation

Abstract

The ESPERANTO project aims at pushing speech processing technologies to their next step in order to enable the diffusion of these technologies in European SMEs and to maximize and securize their use in the civil society for forensic, health or education. The ESPERANTO consortium forsees that the next generation of artificial intelligence algorithms for speech processing should : 1. be more accessible : via a larger number of spoken languages, and for applications where resources are strongly limited (health, education, robotics); 2. integrate a human in the loop to guaranty a higher usability and ease of deployment and maintenance; 3. be explainable in order to enable sensitive applications related to forensic or health and contribute to personal data preservation by detecting and characterizing existing biases due to the data-driven nature of current speech technologies. ESPERANTO intends to lead the scientific community by releasing evaluation metrics, protocols and standards that will boost the development and evaluation of this new generation of algorithms. To achieve this ambitious goal, the ESPERANTO project gathers a large and trans-sectorial community of experts in speech related applications such as speech transcription, separation, enhancement, translation, understanding and speaker recognition and diarization to transfer knowledge, organize, produce and standardize resources with the aim of catalyzing and cross-pollenizing this area. The main goals of the ESPERANTO project are: - support the development of open-source tools that will encourage fast developement, exchanges and reproducibility; - produce tutorials and competitive baselines on various topics of speech processing in order to boost the fostering of new speech-AI students, researchers and engineers; - facilitate the collection and sharing of linguistic and speech resources through standards; - organize workshops to progress on the speech technologies and favor tranfer of knowledge.

Team members
Matějka Pavel, Ing., Ph.D. (UPGM FIT VUT) , research leader
Burget Lukáš, doc. Ing., Ph.D. (UPGM FIT VUT) , team leader
Plchot Oldřich, Ing., Ph.D. (UPGM FIT VUT) , team leader
Rohdin Johan A., Dr. (UPGM FIT VUT) , team leader
Kohlová Renata, Ing. (UPGM FIT VUT)
Landini Federico Nicolás (UPGM FIT VUT)
Mošner Ladislav, Ing. (UPGM FIT VUT)
Silnova Anna, MSc., Ph.D. (UPGM FIT VUT)
Publications

2022

  • SILNOVA Anna, STAFYLAKIS Themos, MOŠNER Ladislav, PLCHOT Oldřich, ROHDIN Johan A., MATĚJKA Pavel, BURGET Lukáš, GLEMBEK Ondřej and BRUMMER Johan Nikolaas Langenhoven. Analyzing speaker verification embedding extractors and back-ends under language and channel mismatch. In: Proceedings of The Speaker and Language Recognition Workshop (Odyssey 2022). Beijing: International Speech Communication Association, 2022, pp. 9-16. Detail
  • KOCOUR Martin, UMESH Jahnavi, KARAFIÁT Martin, ŠVEC Ján, LOPEZ Fernando, BENEŠ Karel, DIEZ Sánchez Mireia, SZŐKE Igor, LUQUE Jordi, VESELÝ Karel, BURGET Lukáš and ČERNOCKÝ Jan. BCN2BRNO: ASR System Fusion for Albayzin 2022 Speech to Text Challenge. In: Proceedings of IberSpeech 2022. Granada: International Speech Communication Association, 2022, pp. 276-280. Detail
  • ALAM Jahangir, BURGET Lukáš, GLEMBEK Ondřej, MATĚJKA Pavel, MOŠNER Ladislav, PLCHOT Oldřich, ROHDIN Johan A., SILNOVA Anna and STAFYLAKIS Themos et al. Development of ABC systems for the 2021 edition of NIST Speaker Recognition evaluation. In: Proceedings of The Speaker and Language Recognition Workshop (Odyssey 2022). Beijing: International Speech Communication Association, 2022, pp. 346-353. Detail
  • PENG Junyi, GU Rongzhi, MOŠNER Ladislav, PLCHOT Oldřich, BURGET Lukáš and ČERNOCKÝ Jan. Learnable Sparse Filterbank for Speaker Verification. In: Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. Incheon: International Speech Communication Association, 2022, pp. 5110-5114. ISSN 1990-9772. Detail
  • MOŠNER Ladislav, PLCHOT Oldřich, BURGET Lukáš and ČERNOCKÝ Jan. Multi-Channel Speaker Verification with Conv-Tasnet Based Beamformer. In: ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. Singapore: IEEE Signal Processing Society, 2022, pp. 7982-7986. ISBN 978-1-6654-0540-9. Detail
  • MOŠNER Ladislav, PLCHOT Oldřich, BURGET Lukáš and ČERNOCKÝ Jan. Multisv: Dataset for Far-Field Multi-Channel Speaker Verification. In: ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. Singapore: IEEE Signal Processing Society, 2022, pp. 7977-7981. ISBN 978-1-6654-0540-9. Detail
  • BRUMMER Johan Nikolaas Langenhoven, SWART Albert du Preez, MOŠNER Ladislav, SILNOVA Anna, PLCHOT Oldřich, STAFYLAKIS Themos and BURGET Lukáš. Probabilistic Spherical Discriminant Analysis: An Alternative to PLDA for length-normalized embeddings. In: Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. Incheon: International Speech Communication Association, 2022, pp. 1446-1450. ISSN 1990-9772. Detail
  • STAFYLAKIS Themos, MOŠNER Ladislav, PLCHOT Oldřich, ROHDIN Johan A., SILNOVA Anna, BURGET Lukáš and ČERNOCKÝ Jan. Training Speaker Embedding Extractors Using Multi-Speaker Audio with Unknown Speaker Boundaries. In: Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. Incheon: International Speech Communication Association, 2022, pp. 605-609. ISSN 1990-9772. Detail

2021

  • STAFYLAKIS Themos, ROHDIN Johan A. and BURGET Lukáš. Speaker embeddings by modeling channel-wise correlations. In: Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. Brno: International Speech Communication Association, 2021, pp. 501-505. ISSN 1990-9772. Detail
Back to top