Towards improving ASR robustness for PSN and GSM telephone applications

Mokbel, Chafic; Mauuary, Laurent; Karray, Lamia; Jouvet, Denis; Monné, Jean; Simonin, Jacques; Bartkova, Katarina

doi:10.1016/s0167-6393(97)00042-3

Cited by 34 publications

(25 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…When both clean and GSM data are used during training in``multistyle'' fashion, there is a further reduction of word error rate of 0.3%. Some authors have noticed improvement when performing multistyle training in cellular and telephone communications or by judicious use of both types of models (Puel and Obrecht, 1997;Das et al, 1999;Mokbel et al, 1996). In our case, the same utterances that were used for training the clean models were coded and reused in the multistyle training.…”

Section: Recognition System Setup and Baseline Experimentsmentioning

confidence: 96%

“…While functionally similar to that of their public-switched telephone network (PSTN) counterparts, speech recognition systems and applications that include a wireless link with coding such as GSM confront the end user with the additional potential problems of wireless-channel noise, data dropouts, and signal degradation originated by the speech codec. Mobile applications provide the additional problem of potentially more varied and more intense environmental noise (Mokbel et al, 1996;HaebUmbach, 1997). These phenomena result in error rates for speech recognition systems which are substantially higher than their counterparts in non-mobile environment (e.g., Das et al, 1999;Delphin-Poulat and Mokbel, 1997;Elvira and Torrecilla, 1998;Mokbel et al, 1996).…”

Section: Introductionmentioning

confidence: 99%

“…Mobile applications provide the additional problem of potentially more varied and more intense environmental noise (Mokbel et al, 1996;HaebUmbach, 1997). These phenomena result in error rates for speech recognition systems which are substantially higher than their counterparts in non-mobile environment (e.g., Das et al, 1999;Delphin-Poulat and Mokbel, 1997;Elvira and Torrecilla, 1998;Mokbel et al, 1996).…”

Section: Introductionmentioning

confidence: 99%

“…Gupta et al (1996) have focused on robustness to environmental noise encountered in mobile applications. Also, there is work regarding the application and tailoring of robustness techniques to the problem, such as spectral subtraction, adaptive ®ltering and model adaptation (Mokbel et al, 1996), robust front ends (Dufour et al, 1996), bias removal and equalization (Delphin-Poulat and Mokbel 1997). Some work has focused on the use of robust acoustic modeling (Puel and Obrecht 1997) and model adaptation (Soulas et al, 1997).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Distortion-class modeling for robust speech recognition under GSM RPE-LTP coding

Huerta

Stern

2001

Speech Communication

View full text Add to dashboard Cite

We present a method to reduce the degradation in recognition accuracy introduced by full-rate GSM RPE-LTP coding by combining sets of acoustic models trained under di erent distortion conditions. During recognition, the a posteriori probabilities of an utterance are calculated as a weighted sum of the posteriors corresponding to the individual models. The phonemes used by the systemÕs word pronunciations are grouped into classes according to amount of distortion they undergo in coding. The acoustic model used in the decoding process is a weighted combination of models derived from clean speech and models derived from speech that had been degraded by GSM coding (the source models), with the relative combination of the two sources depending on the extent to which each class of phonemes is degraded by the coding process. To determine the distortion class membership, and hence the weights, we measure the spectral distortion introduced to the quantized long-term residual by the RPE-LTP codec. We discuss how this distortion varies according to phonetic class. The method described reduces the degradation in recognition accuracy introduced by GSM coding of sentences in the TIMIT database by more than 70% relative to the baseline accuracy obtained in matched training and testing conditions with respect to a system using the source acoustic models, and up to 60% relative to the best baseline systems regardless of the number of Gaussians. Ó

show abstract

Section: Recognition System Setup and Baseline Experimentsmentioning

confidence: 96%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Distortion-class modeling for robust speech recognition under GSM RPE-LTP coding

Huerta

Stern

2001

Speech Communication

View full text Add to dashboard Cite

show abstract

“…The effects of various speech coding algorithms on automatic speech recognition (ASR) performance has been studied by several authors [7], [6], [16], [19]. In [16], [19], [10], [15], [9], severe ASR performance degradation was observed for a distributed wireless speech recognition system, especially in the case of transmission errors that occur in bursts.…”

Section: Introductionmentioning

confidence: 99%

An error-protected speech recognition system for wireless communications

Weerackody¹,

Reichl²,

Potamianos³

2002

IEEE Trans. Wireless Commun.

View full text Add to dashboard Cite

uture wireless multimedia terminals will have a variety of applications that require speech recognition capabilities.In this paper, we consider a robust distributed speech recognition system where representative parameters of the speech signal are extracted at the wireless terminal and transmitted to a centralized automatic speech recognition (ASR) server. We propose several unequal error protection schemes for the ASR bit stream and demonstrate the satisfactory performance of these schemes for typical wireless cellular channels. In addition, a "soft-feature" error concealment strategy is introduced at the ASR server that uses "soft-outputs" from the channel decoder to compute the marginal distribution of only the reliable features during likelihood computation at the speech recognizer. This soft-feature error concealment techniques reduces the ASR error rate by up to four times for certain channels. Also considered is a channel decoding technique with source information that improves ASR performance. uture wireless multimedia terminals will have a variety of applications that require speech recognition capabilities. In this paper, we consider a robust distributed speech recognition system where representative parameters of the speech signal are extracted at the wireless terminal and transmitted to a centralized automatic speech recognition (ASR) server. We propose several unequal error protection schemes for the ASR bit stream and demonstrate the satisfactory performance of these schemes for typical wireless cellular channels. In addition, a "soft-feature" error concealment strategy is introduced at the ASR server that uses "soft-outputs" from the channel decoder to compute the marginal distribution of only the reliable features during likelihood computation at the speech recognizer. This soft-feature error concealment techniques reduces the ASR error rate by up to four times for certain channels. Also considered is a channel decoding technique with source information that improves ASR performance. F

show abstract

A Study on Speech Coders for Automatic Speech Recognition in Adverse Communication Environments

Choi

2011

Informatics Engineering and Information Science

View full text Add to dashboard Cite

Towards improving ASR robustness for PSN and GSM telephone applications

Cited by 34 publications

References 15 publications

Distortion-class modeling for robust speech recognition under GSM RPE-LTP coding

Distortion-class modeling for robust speech recognition under GSM RPE-LTP coding

An error-protected speech recognition system for wireless communications

A Study on Speech Coders for Automatic Speech Recognition in Adverse Communication Environments

Contact Info

Product

Resources

About