DeepAI
Log In Sign Up

Gender domain adaptation for automatic speech recognition task

10/08/2020
by   Sokolov Artem, et al.
0

This paper is focused on the finetuning of acoustic models for speaker adap-tation based on a given gender. We pretrained the Transformer baseline model on Librispeech-960 and conduct experiments with finetuning on the gender-specific test subsets. Our approach leads to 5 the male subset if the layers in the encoder and decoder are not frozen, but the tuning is started from the last checkpoints. Moreover, we adapted our general model on the full L2 Arctic dataset of accented speech and finetuned it for particular speakers and male and female genders separately. The models trained on the gender subsets obtained 1-2 the model tuned on the whole L2 Arctic dataset. Finally, we tested the concatenation of the pretrained x-vector voice embeddings and embeddings from conventional encoder, but its gain in accuracy is not significant.

READ FULL TEXT

page 1

page 2

page 3

page 4

08/04/2021

Unsupervised Domain Adaptation in Speech Recognition using Phonetic Features

Automatic speech recognition is a difficult problem in pattern recogniti...
11/07/2021

Retrieving Speaker Information from Personalized Acoustic Models for Speech Recognition

The widespread of powerful personal devices capable of collecting voice ...
02/18/2022

Domain Adaptation of low-resource Target-Domain models using well-trained ASR Conformer Models

In this paper, we investigate domain adaptation for low-resource Automat...
08/07/2020

Investigation of Speaker-adaptation methods in Transformer based ASR

End-to-end models are fast replacing conventional hybrid models in autom...
05/08/2021

Robustness of end-to-end Automatic Speech Recognition Models – A Case Study using Mozilla DeepSpeech

When evaluating the performance of automatic speech recognition models, ...
11/01/2022

Generating Gender-Ambiguous Text-to-Speech Voices

The gender of a voice assistant or any voice user interface is a central...
07/07/2020

X-vectors: New Quantitative Biomarkers for Early Parkinson's Disease Detection from Speech

Many articles have used voice analysis to detect Parkinson's disease (PD...