Advancing speaker embedding learning: Wespeaker toolkit for research and production

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F24%3APU155129" target="_blank" >RIV/00216305:26230/24:PU155129 - isvavai.cz</a>

Výsledek na webu

<a href="https://pdf.sciencedirectassets.com/271578/1-s2.0-S0167639324X00060/1-s2.0-S0167639324000761/main.pdf?X-Amz-Security-Token=IQoJb3JpZ2luX2VjEAsaCXVzLWVhc3QtMSJIMEYCIQC8Doe66%2Bu6V%2FODd2NY6EZwVTEeN05avzWi09%2FPx3ob%2FQIhAP%2BOyz3L2hXSsDYY4l3zSuz1pzOjFiaTh" target="_blank" >https://pdf.sciencedirectassets.com/271578/1-s2.0-S0167639324X00060/1-s2.0-S0167639324000761/main.pdf?X-Amz-Security-Token=IQoJb3JpZ2luX2VjEAsaCXVzLWVhc3QtMSJIMEYCIQC8Doe66%2Bu6V%2FODd2NY6EZwVTEeN05avzWi09%2FPx3ob%2FQIhAP%2BOyz3L2hXSsDYY4l3zSuz1pzOjFiaTh</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1016/j.specom.2024.103104" target="_blank" >10.1016/j.specom.2024.103104</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Advancing speaker embedding learning: Wespeaker toolkit for research and production

Popis výsledku v původním jazyce

Speaker modeling plays a crucial role in various tasks, and fixed-dimensional vector representations, known as speaker embeddings, are the predominant modeling approach. These embeddings are typically evaluated within the framework of speaker verification, yet their utility extends to a broad scope of related tasks including speaker diarization, speech synthesis, voice conversion, and target speaker extraction. This paper presents Wespeaker, a user-friendly toolkit designed for both research and production purposes, dedicated to the learning of speaker embeddings. Wespeaker offers scalable data management, state-of-the-art speaker embedding models, and self-supervised learning training schemes with the potential to leverage large-scale unlabeled real-world data. The toolkit incorporates structured recipes that have been successfully adopted in winning systems across various speaker verification challenges, ensuring highly competitive results. For production-oriented development, Wespeaker integrates CPU- and GPU-compatible deployment and runtime codes, supporting mainstream platforms such as Windows, Linux, Mac and on-device chips such as horizon X3'PI. Wespeaker also provides off-the-shelf high-quality speaker embeddings by providing various pretrained models, which can be effortlessly applied to different tasks that require speaker modeling. The toolkit is publicly available at https://github.com/wenet-e2e/wespeaker.

Název v anglickém jazyce

Advancing speaker embedding learning: Wespeaker toolkit for research and production

Popis výsledku anglicky

Speaker modeling plays a crucial role in various tasks, and fixed-dimensional vector representations, known as speaker embeddings, are the predominant modeling approach. These embeddings are typically evaluated within the framework of speaker verification, yet their utility extends to a broad scope of related tasks including speaker diarization, speech synthesis, voice conversion, and target speaker extraction. This paper presents Wespeaker, a user-friendly toolkit designed for both research and production purposes, dedicated to the learning of speaker embeddings. Wespeaker offers scalable data management, state-of-the-art speaker embedding models, and self-supervised learning training schemes with the potential to leverage large-scale unlabeled real-world data. The toolkit incorporates structured recipes that have been successfully adopted in winning systems across various speaker verification challenges, ensuring highly competitive results. For production-oriented development, Wespeaker integrates CPU- and GPU-compatible deployment and runtime codes, supporting mainstream platforms such as Windows, Linux, Mac and on-device chips such as horizon X3'PI. Wespeaker also provides off-the-shelf high-quality speaker embeddings by providing various pretrained models, which can be effortlessly applied to different tasks that require speaker modeling. The toolkit is publicly available at https://github.com/wenet-e2e/wespeaker.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/VB02000060" target="_blank" >VB02000060: Nástroje boje proti hlasovým DeepFakes</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Speech Communication

ISSN

0167-6393

e-ISSN

1872-7182

Svazek periodika

162

Číslo periodika v rámci svazku

103104

Stát vydavatele periodika

NL - Nizozemsko

Počet stran výsledku

12

Strana od-do

1-12

Kód UT WoS článku

001279201500001

EID výsledku v databázi Scopus

2-s2.0-85199203394