Recognition of Dysarthric Speech Using Voice Parameters for Speaker Adaptation and Multi-Taper Spectral Estimation

Bhat, Chitralekha; Vachhani, Bhavik; Kopparapu, Sunil Kumar

doi:10.21437/interspeech.2016-1085

Cited by 17 publications

(14 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several videos available on the internet were selected and visual subtitles were generated (Bhat et al, 2013b) as described earlier.…”

Section: Methodsmentioning

confidence: 99%

Usable technology for the differently abled-hearing impairment

Bhat¹,

Kopparapu²

2018

IJHT

Self Cite

View full text Add to dashboard Cite

Abstract:Hearing loss poses a serious challenge to access information that are in the form of audio or video on the internet. Text subtitles or closed captioning is a popular mechanism to make the audio embedded in the videos accessible to the hearing impaired. Alternatively, a sign language interpreter may translate the audio into sign language gestures. However, these aids are either expensive or technologically infeasible. In this paper, we propose a technique to enable a person with hearing impairment see the audio. The novel assistive tool is envisioned as a visual aid for the hearing impaired to comprehend the audio contained in a video. The tool produces a speech synchronised visual lip movement sequence for a video. The lip movement sequence is then superimposed onto and displayed along with the original video. A number of experiments have been conducted to determine the feasibility and usefulness of the proposed visual aid.

show abstract

“…Several videos available on the internet were selected and visual subtitles were generated (Bhat et al, 2013b) as described earlier.…”

Section: Methodsmentioning

confidence: 99%

Usable technology for the differently abled-hearing impairment

Bhat¹,

Kopparapu²

2018

IJHT

Self Cite

View full text Add to dashboard Cite

show abstract

“…The main contributions of the paper are summarized below: 1) To the best of our knowledge, our novel spectro-temporal deep feature based adaptation approach is the first work to exploit auxiliary speaker embedding features in disordered speech adaptation. In contrast, prior works [10,12,13,27,[34][35][36][37][38][39] focused on feature transformation and model based adaptation. Speaker embedding features, e.g.…”

Section: Introductionmentioning

confidence: 98%

“…In [12], a combination of MLLR and MAP adaptation were used in speaker adaptive training (SAT) of SI GMM-HMM models. In [36], f-MLLR based SAT was studied. In [37], regularized speaker adaptation on Kullback-Leibler divergencebased HMMs (KL-HMMs) was conducted.…”

Section: Introductionmentioning

confidence: 99%

Spectro-Temporal Deep Features for Disordered Speech Assessment and Recognition

Geng,

Liu,

et al. 2022

Preprint

View full text Add to dashboard Cite

Automatic recognition of disordered speech remains a highly challenging task to date. Sources of variability commonly found in normal speech including accent, age or gender, when further compounded with the underlying causes of speech impairment and varying severity levels, create large diversity among speakers. To this end, speaker adaptation techniques play a vital role in current speech recognition systems. Motivated by the spectro-temporal level differences between disordered and normal speech that systematically manifest in articulatory imprecision, decreased volume and clarity, slower speaking rates and increased dysfluencies, novel spectro-temporal subspace basis embedding deep features derived by SVD decomposition of speech spectrum are proposed to facilitate both accurate speech intelligibility assessment and auxiliary feature based speaker adaptation of state-of-the-art hybrid DNN and end-to-end disordered speech recognition systems. Experiments conducted on the UASpeech corpus suggest the proposed spectro-temporal deep feature adapted systems consistently outperformed baseline i-Vector adaptation by up to 2.63% absolute (8.6% relative) reduction in word error rate (WER) with or without data augmentation. Learning hidden unit contribution (LHUC) based speaker adaptation was further applied. The final speaker adapted system using the proposed spectral basis embedding features gave an overall WER of 25.6% on the UASpeech test set of 16 dysarthric speakers.

show abstract

“…A comparative study of several types of ASR sys-tems including maximum likelihood and maximum a posteriori (MAP) adaptation showed a significant improvement in dysarthric speech recognition when speaker adaptation using MAP adaptation was applied [6]. Word error rate for dysarthric speech was reduced using voice parameters such as jitter and shimmer along with multi-taper Mel-frequency Cepstral Coefficients (MFCC) followed by speaker adaptation [7], and using Elman back-propagation network (EBN) which is a recurrent, self supervised neural network along with glottal features and MFCC in [8]. A multi-stage deep neural network (DNN) training scheme is used to better model dysarthric speech, wherein only a small amount of in-domain training data showed considerable improvement in the recognition of dysarthric speech [9].…”

Section: Introductionmentioning

confidence: 99%

Dysarthric Speech Recognition Using Time-delay Neural Network Based Denoising Autoencoder

et al. 2018

Self Cite

View full text Add to dashboard Cite

Dysarthria is a manisfestation of the disruption in the neuromuscular physiology resulting in uneven, slow, slurred, harsh or quiet speech. Dysarthric speech poses serious challenges to automatic speech recognition, considering this speech is difficult to decipher for both humans and machines. The objective of this work is to enhance dysarthric speech features to match that of healthy control speech. We use a Time-Delay Neural Network based Denoising Autoencoder (TDNN-DAE) to enhance the dysarthric speech features. The dysarthric speech thus enhanced is recognized using a DNN-HMM based Automatic Speech Recognition (ASR) engine. This methodology was evaluated for speaker-independent (SI) and speaker-adapted (SA) systems. Absolute improvements of 13% and 3% was observed in the ASR performance for SI and SA systems respectively as compared with unenhanced dysarthric speech recognition.

show abstract

Recognition of Dysarthric Speech Using Voice Parameters for Speaker Adaptation and Multi-Taper Spectral Estimation

Cited by 17 publications

References 29 publications

Usable technology for the differently abled-hearing impairment

Usable technology for the differently abled-hearing impairment

Spectro-Temporal Deep Features for Disordered Speech Assessment and Recognition

Dysarthric Speech Recognition Using Time-delay Neural Network Based Denoising Autoencoder

Contact Info

Product

Resources

About