1 |
LeBenchmark: A Reproducible Framework for Assessing Self-Supervised Representation Learning from Speech
|
|
Evain, Solène; Nguyen, Ha; Le, Hang; Zanon Boito, Marcely; Mdhaffar, Salima; Alisamir, Sina; Tong, Ziyi; Tomashenko, Natalia; Dinarelli, Marco; Parcollet, Titouan; Allauzen, Alexandre; Estève, Yannick; Lecouteux, Benjamin; Portet, François; Rossato, Solange; Ringeval, Fabien; Schwab, Didier; Besacier, Laurent
|
|
In: INTERSPEECH 2021: Conference of the International Speech Communication Association ; https://hal.archives-ouvertes.fr/hal-03317730 ; INTERSPEECH 2021: Conference of the International Speech Communication Association, Aug 2021, Brno, Czech Republic (2021)
|
|
Abstract:
International audience ; Self-Supervised Learning (SSL) using huge unlabeled data has been successfully explored for image and natural language processing. Recent works also investigated SSL from speech. They were notably successful to improve performance on downstream tasks such as automatic speech recognition (ASR). While these works suggest it is possible to reduce dependence on labeled data for building efficient speech systems, their evaluation was mostly made on ASR and using multiple and heterogeneous experimental settings (most of them for English). This questions the objective comparison of SSL approaches and the evaluation of their impact on building speech systems. In this paper, we propose LeBenchmark: a reproducible framework for assessing SSL from speech. It not only includes ASR (high and low resource) tasks but also spoken language understanding, speech translation and emotion recognition. We also focus on speech technologies in a language different than English: French. SSL models of different sizes are trained from carefully sourced and documented datasets. Experiments show that SSL is beneficial for most but not all tasks which confirms the need for exhaustive and reliable benchmarks to evaluate its real impact. LeBenchmark is shared with the scientific community for reproducible research in SSL from speech.
|
|
Keyword:
[INFO.INFO-AI]Computer Science [cs]/Artificial Intelligence [cs.AI]; ASR; Automatic Emotion Recognition; Self-Supervised Representation Learning; SLU; Speech Translation
|
|
URL: https://hal.archives-ouvertes.fr/hal-03317730/file/FLOWBERT_IS2021.pdf https://hal.archives-ouvertes.fr/hal-03317730/document https://hal.archives-ouvertes.fr/hal-03317730
|
|
BASE
|
|
Hide details
|
|
2 |
LeBenchmark: A Reproducible Framework for Assessing Self-Supervised Representation Learning from Speech
|
|
|
|
In: INTERSPEECH 2021: ; INTERSPEECH 2021: Conference of the International Speech Communication Association ; https://hal.archives-ouvertes.fr/hal-03317730 ; INTERSPEECH 2021: Conference of the International Speech Communication Association, Aug 2021, Brno, Czech Republic (2021)
|
|
BASE
|
|
Show details
|
|
3 |
LeBenchmark: A Reproducible Framework for Assessing Self-Supervised Representation Learning from Speech
|
|
|
|
In: INTERSPEECH 2021: ; INTERSPEECH 2021: Conference of the International Speech Communication Association ; https://hal.archives-ouvertes.fr/hal-03317730 ; INTERSPEECH 2021: Conference of the International Speech Communication Association, Aug 2021, Brno, Czech Republic (2021)
|
|
BASE
|
|
Show details
|
|
4 |
TArC : Incrementally and Semi-Automatically Collecting a Tunisian arabish Corpus ; TArC. Un corpus d'arabish tunisien
|
|
|
|
In: Actes de la 6e conférence conjointe Journées d'Études sur la Parole (JEP, 33e édition), Traitement Automatique des Langues Naturelles (TALN, 27e édition), Rencontre des Étudiants Chercheurs en Informatique pour le Traitement Automatique des Langues (RÉCITAL, 22e édition). Volume 2 : Traitement Automatique des Langues Naturelles ; 6e conférence conjointe Journées d'Études sur la Parole (JEP, 33e édition), Traitement Automatique des Langues Naturelles (TALN, 27e édition), Rencontre des Étudiants Chercheurs en Informatique pour le Traitement Automatique des Langues (RÉCITAL, 22e édition). Volume 2 : Traitement Automatique des Langues Naturelles ; https://hal.archives-ouvertes.fr/hal-02784772 ; 6e conférence conjointe Journées d'Études sur la Parole (JEP, 33e édition), Traitement Automatique des Langues Naturelles (TALN, 27e édition), Rencontre des Étudiants Chercheurs en Informatique pour le Traitement Automatique des Langues (RÉCITAL, 22e édition). Volume 2 : Traitement Automatique des Langues Naturelles, Jun 2020, Nancy, France. pp.232-240 ; https://jep-taln2020.loria.fr/ (2020)
|
|
BASE
|
|
Show details
|
|
5 |
TArC Un corpus d'arabish tunisien
|
|
|
|
In: Traitement Automatique des Langues Naturelles (TALN) ; https://hal.archives-ouvertes.fr/hal-03096013 ; Traitement Automatique des Langues Naturelles (TALN), Jun 2020, Nancy, France (2020)
|
|
BASE
|
|
Show details
|
|
6 |
TArC: Incrementally and Semi-Automatically Collecting a Tunisian Arabish Corpus
|
|
|
|
In: Language Resources and Evaluation Conference (LREC) ; Proceedings of the 12th Language Resources and Evaluation Conference, 2020 ; https://hal.archives-ouvertes.fr/hal-03095971 ; Proceedings of the 12th Language Resources and Evaluation Conference, 2020, Jun 2020, Marseille, France (2020)
|
|
BASE
|
|
Show details
|
|
7 |
A Data-Efficient End-to-End Spoken Language Understanding Architecture
|
|
|
|
In: International Conference on Acoustics, Speech, and Signal Processing (ICASSP) ; https://hal.archives-ouvertes.fr/hal-03094850 ; International Conference on Acoustics, Speech, and Signal Processing (ICASSP), May 2020, Barcellone, Spain (2020)
|
|
BASE
|
|
Show details
|
|
8 |
A Data Efficient End-To-End Spoken Language Understanding Architecture ...
|
|
|
|
BASE
|
|
Show details
|
|
9 |
TArC: Incrementally and Semi-Automatically Collecting a Tunisian Arabish Corpus ...
|
|
|
|
BASE
|
|
Show details
|
|
10 |
Seq2Biseq: Bidirectional Output-wise Recurrent Neural Networks for Sequence Modelling
|
|
|
|
In: CICLing 2019 - 20th International Conference on Computational Linguistics and Intelligent Text Processing ; https://hal.inria.fr/hal-02085093 ; CICLing 2019 - 20th International Conference on Computational Linguistics and Intelligent Text Processing, Apr 2019, La Rochelle, France (2019)
|
|
BASE
|
|
Show details
|
|
11 |
ANCOR-AS: Enriching the ANCOR Corpus with Syntactic Annotations
|
|
|
|
In: LREC 2018 - 11th edition of the Language Resources and Evaluation Conference ; https://hal.inria.fr/hal-01744572 ; LREC 2018 - 11th edition of the Language Resources and Evaluation Conference, May 2018, Miyazaki, Japan ; http://lrec2018.lrec-conf.org/en/ (2018)
|
|
BASE
|
|
Show details
|
|
12 |
Coreference Resolution for French Oral Data: Machine Learning Experiments with ANCOR
|
|
|
|
In: Computational Linguistics and Intelligent Text Processing. ; https://hal.archives-ouvertes.fr/hal-01889593 ; Computational Linguistics and Intelligent Text Processing., n° 9623-9624, Springer, 2018, Lecture Notes in Computer Science (2018)
|
|
BASE
|
|
Show details
|
|
13 |
Modeling a label global context for sequence tagging in recurrent neural networks ; Modélisation d'un contexte global d'étiquettes pour l'étiquetage de séquences dans les réseaux neuronaux récurrents
|
|
|
|
In: Journée commune AFIA-ATALA sur le Traitement Automatique des Langues et l’Intelligence Artificielle pendant la onzième édition de la plate-forme Intelligence Artificielle (PFIA 2018) ; https://hal.archives-ouvertes.fr/hal-02002111 ; Journée commune AFIA-ATALA sur le Traitement Automatique des Langues et l’Intelligence Artificielle pendant la onzième édition de la plate-forme Intelligence Artificielle (PFIA 2018), Jul 2018, Nancy, France ; https://pfia2018.loria.fr/journee-tal/ (2018)
|
|
BASE
|
|
Show details
|
|
14 |
Experiences in using deep and shallow parsing to detect entity mentions in oral French ; Apports des analyses syntaxiques pour la détection automatique de mentions dans un corpus de français oral
|
|
|
|
In: TALN 2017 ; https://hal.inria.fr/hal-01558711 ; TALN 2017, Association pour le Traitement Automatique des Langues (ATALA), Jun 2017, Orléans, France ; http://taln2017.cnrs.fr (2017)
|
|
BASE
|
|
Show details
|
|
15 |
Structured Named Entity Recognition by Cascading CRFs
|
|
|
|
In: Intelligent Text Processing and Computational Linguistics (CICling) ; https://hal.archives-ouvertes.fr/hal-01579109 ; Intelligent Text Processing and Computational Linguistics (CICling), Apr 2017, Budapest, Hungary (2017)
|
|
BASE
|
|
Show details
|
|
16 |
Label-Dependencies Aware Recurrent Neural Networks
|
|
|
|
In: Intelligent Text Processing and Computational Linguistics (CICling) ; https://hal.archives-ouvertes.fr/hal-01579071 ; Intelligent Text Processing and Computational Linguistics (CICling), Apr 2017, Budapest, Hungary ; http://www.cicling.org/2017/ (2017)
|
|
BASE
|
|
Show details
|
|
17 |
Understanding Social Media Texts with Minimum Human Effort on #Twitter
|
|
|
|
In: Language and the new (instant) media (PLIN) ; https://hal.archives-ouvertes.fr/hal-01490018 ; Language and the new (instant) media (PLIN), May 2016, Louvain-la-Neuve, Belgium (2016)
|
|
BASE
|
|
Show details
|
|
18 |
Coreference Resolution for French Oral Data: Machine Learning Experiments with ANCOR
|
|
|
|
In: 17th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing'2016) ; https://hal.archives-ouvertes.fr/hal-01344977 ; 17th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing'2016), Apr 2016, Konya, Turkey ; http://www.cicling.org/2016/ (2016)
|
|
BASE
|
|
Show details
|
|
19 |
Domain Adaptation for Named Entity Recognition Using CRFs
|
|
|
|
In: LREC 2016 ; https://hal.archives-ouvertes.fr/hal-01473397 ; LREC 2016, May 2016, Portoroz, Slovenia ; http://lrec2016.lrec-conf.org/en/ (2016)
|
|
BASE
|
|
Show details
|
|
20 |
New Recurrent Neural Network Variants for Sequence Labeling
|
|
|
|
In: ISSN: 0302-9743 ; Lecture Notes in Computer Science ; 17th International Conference on Intelligent Text Processing and Computational Linguistics ; https://hal.archives-ouvertes.fr/hal-01489955 ; 17th International Conference on Intelligent Text Processing and Computational Linguistics, Apr 2016, Konya, Turkey (2016)
|
|
BASE
|
|
Show details
|
|
|
|