Auxiliary Loss Function for Target Speech Extraction and Recognition with Weak Supervision Based on Speaker Characteristics

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F21%3APU142960" target="_blank" >RIV/00216305:26230/21:PU142960 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.isca-speech.org/archive/interspeech_2021/zmolikova21_interspeech.html" target="_blank" >https://www.isca-speech.org/archive/interspeech_2021/zmolikova21_interspeech.html</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.21437/Interspeech.2021-986" target="_blank" >10.21437/Interspeech.2021-986</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Auxiliary Loss Function for Target Speech Extraction and Recognition with Weak Supervision Based on Speaker Characteristics

Popis výsledku v původním jazyce

Automatic speech recognition systems deteriorate in presence of overlapped speech. A popular approach to alleviate this is target speech extraction. The extraction system is usually trained with a loss function measuring the discrepancy between the estimated and the reference target speech. This often leads to distortions to the target signal which is detrimental to the recognition accuracy. Additionally, it is necessary to have the strong supervision provided by parallel data consisting of speech mixtures and single-speaker signals. We propose an auxiliary loss function for retraining the target speech extraction. It is composed of two parts: first, a speaker identity loss, forcing the estimated speech to have correct speaker characteristics, and second, a mixture consistency loss, making the extracted sources sum back to the original mixture. The only supervision required for the proposed loss is speaker characteristics obtained from several segments spoken by the target speaker. Such weak supervision makes the loss suitable for adapting the system directly on real recordings. We show that the proposed loss yields signals more suitable for speech recognition and further, we can gain additional improvements by adaptation to target data. Overall, we can reduce the word error rate on LibriCSS dataset from 27.4% to 24.0%.

Název v anglickém jazyce

Auxiliary Loss Function for Target Speech Extraction and Recognition with Weak Supervision Based on Speaker Characteristics

Popis výsledku anglicky

Automatic speech recognition systems deteriorate in presence of overlapped speech. A popular approach to alleviate this is target speech extraction. The extraction system is usually trained with a loss function measuring the discrepancy between the estimated and the reference target speech. This often leads to distortions to the target signal which is detrimental to the recognition accuracy. Additionally, it is necessary to have the strong supervision provided by parallel data consisting of speech mixtures and single-speaker signals. We propose an auxiliary loss function for retraining the target speech extraction. It is composed of two parts: first, a speaker identity loss, forcing the estimated speech to have correct speaker characteristics, and second, a mixture consistency loss, making the extracted sources sum back to the original mixture. The only supervision required for the proposed loss is speaker characteristics obtained from several segments spoken by the target speaker. Such weak supervision makes the loss suitable for adapting the system directly on real recordings. We show that the proposed loss yields signals more suitable for speech recognition and further, we can gain additional improvements by adaptation to target data. Overall, we can reduce the word error rate on LibriCSS dataset from 27.4% to 24.0%.

Druh

D - Stať ve sborníku

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/LTAIN19087" target="_blank" >LTAIN19087: Multi-lingualita v řečových technologiích</a><br>

Návaznosti

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

Rok uplatnění

2021

Kód důvěrnosti údajů

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

Název statě ve sborníku

Proceedings of 2021 Interspeech

ISBN

—

ISSN

1990-9772

e-ISSN

—

Počet stran výsledku

5

Strana od-do

1464-1468

Název nakladatele

International Speech Communication Association

Místo vydání

Brno

Místo konání akce

Brno

Datum konání akce

30. 8. 2021

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—