On Practical Aspects of Multi-condition Training Based on Augmentation for Reverberation-/Noise-Robust Speech Recognition

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F46747885%3A24220%2F19%3A00007162" target="_blank" >RIV/46747885:24220/19:00007162 - isvavai.cz</a>

Výsledek na webu

<a href="http://dx.doi.org/10.1007/978-3-030-27947-9_21" target="_blank" >http://dx.doi.org/10.1007/978-3-030-27947-9_21</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-030-27947-9_21" target="_blank" >10.1007/978-3-030-27947-9_21</a>

Jazyk výsledku

angličtina

Název v původním jazyce

On Practical Aspects of Multi-condition Training Based on Augmentation for Reverberation-/Noise-Robust Speech Recognition

Popis výsledku v původním jazyce

Multi-condition training achieved through data augmentation belongs to the most successful techniques for noise/reverberation-robust automatic speech recognition (ASR). Its basic principle, i.e., generation of artificially distorted speech signals, is well documented in the literature. However, the specific choice of hyper-parameters for the generation process and its influence on the results of the subsequent ASR is usually not discussed in detail. Often, it is simply assumed that the augmentation should include as many acoustic conditions as possible. When designed in this broad manner, the computational/storage demands of the augmentation process grow rapidly. In this paper, we rather aim for careful selection of a limited number of acoustic conditions that are highly relevant with respect to the target environment. In this manner, we keep the computational requirements feasible, while retaining the improved accuracy of the augmented models. We experimentally analyze two augmentation scenarios and draw conclusions regarding suitable setup choices. The first case concerns augmentation for reverberation-robust ASR. We propose to exploit Clarity C50 as a feature for selection of Room Impulse Responses (RIRs) crucial for the augmentation. We show that mismatches in other RIR-related parameters, such as Reverberation Time T60 or the room dimension, have small influence on ASR accuracy, as long as the training-test conditions are matched from the C50 perspective. Subsequently, we investigate the augmentation for noise-reverberation-robust ASR. We discuss selection of Signal-to-Noise Ratio (SNR), the type of noise and reverberation level of speech. We observe the influence of mismatches in these parameters on the ASR accuracy

Název v anglickém jazyce

On Practical Aspects of Multi-condition Training Based on Augmentation for Reverberation-/Noise-Robust Speech Recognition

Popis výsledku anglicky

Multi-condition training achieved through data augmentation belongs to the most successful techniques for noise/reverberation-robust automatic speech recognition (ASR). Its basic principle, i.e., generation of artificially distorted speech signals, is well documented in the literature. However, the specific choice of hyper-parameters for the generation process and its influence on the results of the subsequent ASR is usually not discussed in detail. Often, it is simply assumed that the augmentation should include as many acoustic conditions as possible. When designed in this broad manner, the computational/storage demands of the augmentation process grow rapidly. In this paper, we rather aim for careful selection of a limited number of acoustic conditions that are highly relevant with respect to the target environment. In this manner, we keep the computational requirements feasible, while retaining the improved accuracy of the augmented models. We experimentally analyze two augmentation scenarios and draw conclusions regarding suitable setup choices. The first case concerns augmentation for reverberation-robust ASR. We propose to exploit Clarity C50 as a feature for selection of Room Impulse Responses (RIRs) crucial for the augmentation. We show that mismatches in other RIR-related parameters, such as Reverberation Time T60 or the room dimension, have small influence on ASR accuracy, as long as the training-test conditions are matched from the C50 perspective. Subsequently, we investigate the augmentation for noise-reverberation-robust ASR. We discuss selection of Signal-to-Noise Ratio (SNR), the type of noise and reverberation level of speech. We observe the influence of mismatches in these parameters on the ASR accuracy

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/TH03010018" target="_blank" >TH03010018: DeepSpot - Multilingvální technologie pro detekci a včasné upozornění</a><br>

Návaznosti

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

Rok uplatnění

2019

Kód důvěrnosti údajů

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

Název statě ve sborníku

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ISBN

978-303027946-2

ISSN

03029743

e-ISSN

—

Počet stran výsledku

13

Strana od-do

251-263

Název nakladatele

Springer Nature Switzerland AG.

Místo vydání

Switzerland

Místo konání akce

Ljubljana; Slovenia

Datum konání akce

1. 1. 2019

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—