FAIRSEQ S2: A Scalable and Integrable Speech Synthesis Toolkit

Changhan Wang; Wei Ning Hsu; Yossi Adi; Adam Polyak; Ann Lee; Peng Jen Chen; Jiatao Gu; Juan Pino

FAIRSEQ S²: A Scalable and Integrable Speech Synthesis Toolkit

Changhan Wang, Wei Ning Hsu, Yossi Adi, Adam Polyak, Ann Lee, Peng Jen Chen, Jiatao Gu, Juan Pino

Meta

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

12 Scopus citations

Abstract

This paper presents FAIRSEQ S², a FAIRSEQ extension for speech synthesis. We implement a number of autoregressive (AR) and non-AR text-to-speech models, and their multi-speaker variants. To enable training speech synthesis models with less curated data, a number of preprocessing tools are built and their importance is shown empirically. To facilitate faster iteration of development and analysis, a suite of automatic metrics is included. Apart from the features added specifically for this extension, FAIRSEQ S² also benefits from the scalability offered by FAIRSEQ and can be easily integrated with other state-of-the-art systems provided in this framework. The code, documentation, and pre-trained models will be made available at https://github.com/pytorch/fairseq/tree/master/examples/speech_synthesis.

Original language	English
Title of host publication	EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing
Subtitle of host publication	System Demonstrations
Publisher	Association for Computational Linguistics (ACL)
Pages	143-152
Number of pages	10
ISBN (Electronic)	9781955917117
State	Published - 2021
Externally published	Yes
Event	2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021 - Virtual, Punta Cana, Dominican Republic Duration: 7 Nov 2021 → 11 Nov 2021

Publication series

Name	EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations

Conference

Conference	2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021
Country/Territory	Dominican Republic
City	Virtual, Punta Cana
Period	7/11/21 → 11/11/21

Bibliographical note

Publisher Copyright:
© 2021 Association for Computational Linguistics.

Cite this

Wang, C., Hsu, W. N., Adi, Y., Polyak, A., Lee, A., Chen, P. J., Gu, J., & Pino, J. (2021). FAIRSEQ S²: A Scalable and Integrable Speech Synthesis Toolkit. In EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations (pp. 143-152). (EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations). Association for Computational Linguistics (ACL).

Wang, Changhan ; Hsu, Wei Ning ; Adi, Yossi et al. / FAIRSEQ S² : A Scalable and Integrable Speech Synthesis Toolkit. EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. Association for Computational Linguistics (ACL), 2021. pp. 143-152 (EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations).

@inproceedings{195d3ad7bfba4316b28d377a23b270c1,

title = "FAIRSEQ S2: A Scalable and Integrable Speech Synthesis Toolkit",

abstract = "This paper presents FAIRSEQ S2, a FAIRSEQ extension for speech synthesis. We implement a number of autoregressive (AR) and non-AR text-to-speech models, and their multi-speaker variants. To enable training speech synthesis models with less curated data, a number of preprocessing tools are built and their importance is shown empirically. To facilitate faster iteration of development and analysis, a suite of automatic metrics is included. Apart from the features added specifically for this extension, FAIRSEQ S2 also benefits from the scalability offered by FAIRSEQ and can be easily integrated with other state-of-the-art systems provided in this framework. The code, documentation, and pre-trained models will be made available at https://github.com/pytorch/fairseq/tree/master/examples/speech_synthesis.",

author = "Changhan Wang and Hsu, {Wei Ning} and Yossi Adi and Adam Polyak and Ann Lee and Chen, {Peng Jen} and Jiatao Gu and Juan Pino",

note = "Publisher Copyright: {\textcopyright} 2021 Association for Computational Linguistics.; 2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021 ; Conference date: 07-11-2021 Through 11-11-2021",

year = "2021",

language = "אנגלית",

series = "EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",

publisher = "Association for Computational Linguistics (ACL)",

pages = "143--152",

booktitle = "EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing",

address = "ארצות הברית",

}

Wang, C, Hsu, WN, Adi, Y, Polyak, A, Lee, A, Chen, PJ, Gu, J & Pino, J 2021, FAIRSEQ S²: A Scalable and Integrable Speech Synthesis Toolkit. in EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, Association for Computational Linguistics (ACL), pp. 143-152, 2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021, Virtual, Punta Cana, Dominican Republic, 7/11/21.

FAIRSEQ S²: A Scalable and Integrable Speech Synthesis Toolkit. / Wang, Changhan; Hsu, Wei Ning; Adi, Yossi et al.
EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. Association for Computational Linguistics (ACL), 2021. p. 143-152 (EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - FAIRSEQ S2

T2 - 2021 Conference on Empirical Methods in Natural Language Processing, EMNLP 2021

AU - Wang, Changhan

AU - Hsu, Wei Ning

AU - Adi, Yossi

AU - Polyak, Adam

AU - Lee, Ann

AU - Chen, Peng Jen

AU - Gu, Jiatao

AU - Pino, Juan

PY - 2021

Y1 - 2021

N2 - This paper presents FAIRSEQ S2, a FAIRSEQ extension for speech synthesis. We implement a number of autoregressive (AR) and non-AR text-to-speech models, and their multi-speaker variants. To enable training speech synthesis models with less curated data, a number of preprocessing tools are built and their importance is shown empirically. To facilitate faster iteration of development and analysis, a suite of automatic metrics is included. Apart from the features added specifically for this extension, FAIRSEQ S2 also benefits from the scalability offered by FAIRSEQ and can be easily integrated with other state-of-the-art systems provided in this framework. The code, documentation, and pre-trained models will be made available at https://github.com/pytorch/fairseq/tree/master/examples/speech_synthesis.

AB - This paper presents FAIRSEQ S2, a FAIRSEQ extension for speech synthesis. We implement a number of autoregressive (AR) and non-AR text-to-speech models, and their multi-speaker variants. To enable training speech synthesis models with less curated data, a number of preprocessing tools are built and their importance is shown empirically. To facilitate faster iteration of development and analysis, a suite of automatic metrics is included. Apart from the features added specifically for this extension, FAIRSEQ S2 also benefits from the scalability offered by FAIRSEQ and can be easily integrated with other state-of-the-art systems provided in this framework. The code, documentation, and pre-trained models will be made available at https://github.com/pytorch/fairseq/tree/master/examples/speech_synthesis.

UR - http://www.scopus.com/inward/record.url?scp=85126960412&partnerID=8YFLogxK

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:85126960412

T3 - EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations

SP - 143

EP - 152

BT - EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing

PB - Association for Computational Linguistics (ACL)

Y2 - 7 November 2021 through 11 November 2021

ER -

Wang C, Hsu WN, Adi Y, Polyak A, Lee A, Chen PJ et al. FAIRSEQ S²: A Scalable and Integrable Speech Synthesis Toolkit. In EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. Association for Computational Linguistics (ACL). 2021. p. 143-152. (EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations).