Localization and selection of speaker-specific information with statistical modeling

Besacier, Laurent; Bonastre, Jean-François; Fredouille, Corinne

doi:10.1016/s0167-6393(99)00070-9

Cited by 46 publications

(50 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This result is also consistent with findings of other studies such as [124,125] that the high frequency region contains the most discriminative speaker information.…”

Section: The Effect Of Speech Bandwidth On Accent Recognitionsupporting

confidence: 93%

Human and computer recognition of regional accents and ethnic groups from British English speech

Hanani

Russell

Carey

2013

Computer Speech & Language

View full text Add to dashboard Cite

Language Recognition Evaluation (LRE) tasks. This system is applied to the problems of regional accent recognition for British English, and ethnic group recognition within a particular accent. We compare the results with human performance and, for accent recognition, the 'text dependent' ACCDIST accent recognition measure. For the fourteen regional accents of British English in the ABI-1 corpus (good quality read speech), our language ID system achieves a recognition accuracy of 86.4%, compared with 95.18% for our best ACCDIST-based system and 58.24% for human listeners. The "Voices across Birmingham" corpus contains significant amounts of telephone conversational speech for the two largest ethnic groups in the city of Birmingham (UK), namely the 'Asian' and 'White' communities. Our language ID system distinguishes between these two groups with an accuracy of 94.3% compared with 90.24% for human listeners. Although direct comparison is difficult, it seems that our language ID system performs much better on the standard twelve class NIST 2003 Language Recognition Evaluation task or the two class ethnic group recognition task than on the fourteen class regional accent recognition task. We conclude that automatic accent recognition is a challenging task for speech technology, and that the use of natural conversational speech may be advantageous for these types of paralinguistic task.One issue with conventional approaches to language ID that use high-order Gaussian Mixture Models (GMMs) and high-dimensional feature vectors is the amount of computing power that they require. Currently, multi-core Graphics Processing Units (GPUs) provide a possible solution at very little cost. In this thesis we also explore the application of GPUs to speech signal and pattern processing, using language ID as a vehicle to demonstrate their benefits. Realisation of the full potential of GPUs requires both effective coding of predetermined algorithms, and, in cases where there is a choice, selection of the algorithm or technique for a specific function that is most able to exploit the properties of the GPU. We demonstrate these principles using the NIST LRE 2003 task, which involves processing over 600 hours of speech. We focus on two parts of the system, namely the acoustic classifier, which is based on a 2048 component GMM, and the acoustic feature extraction process. In the case of the latter we compare a conventional FFT-based analysis with an FIR filter bank, both in terms of their ability to exploit the GPU architecture and language ID performance. With no increase in error rate our GPU based system, with an FIR-based front-end, completes the full NIST LRE 2003 task in 16 hours, compared with 180 hours for the more conventional FFT-based system on a standard CPU (a speed up factor of more than 11).

show abstract

“…This result is also consistent with findings of other studies such as [124,125] that the high frequency region contains the most discriminative speaker information.…”

Section: The Effect Of Speech Bandwidth On Accent Recognitionsupporting

confidence: 93%

Human and computer recognition of regional accents and ethnic groups from British English speech

Hanani

Russell

Carey

2013

Computer Speech & Language

View full text Add to dashboard Cite

show abstract

“…How to identify the reliable feature parts assuming minimum noise information remains a focus of the research for the MF method. Previous studies have suggested different methods (see, for example, [13], [14], [24]- [26]). In this paper, we study the posterior union model (PUM) [27].…”

Section: Combining Wf Nc With Missingfeature Technique (Wf+mf Nc+mf)mentioning

confidence: 99%

“…Other techniques rely on a statistical model of the noise, for example, parallel model combination (PMC) [9], [10], or on the use of microphone arrays [11], [12]. Recent studies on the missing-feature method have shown improved robustness for speech data subjected to partial noise corruption (e.g., [13], [14]). …”

Section: Introductionmentioning

confidence: 99%

A Comparative Study of Methods for Handheld Speaker Verification in Realistic Noisy Conditions

Hazen²,

Glass³

2006

2006 IEEE Odyssey - The Speaker and Language Recognition Workshop

View full text Add to dashboard Cite

A number of methods have been studied and compared for their robustness for speaker verification using noisy speech samples. The methods include Wiener filtering, noise compensation, missing-feature technique, universal compensation, and their combinations. Strategies for combining different techniques are investigated, as a means of further improving noise robustness. A handheld-device database, collected in realistic conditions with noise corruption and transducer mismatch between training and testing, is used in the study. The various techniques and proposed combinations are compared within the same feature and model framework for characterizing the speakers. The experimental results indicate that: 1) usual noise filtering and noise compensation provided very limited robustness to noise corruption, and 2) the proposed technique combinations offered significantly improved noise robustness.

show abstract

“…For this application, the watermark bits are embedded at locations where fewer speaker-specific sub-bands are available. Basically, the discriminative speaker features are contained within the low-and high-frequency bands: the glottis frequency range is between 100 and 400 Hz, the piriform fossa range is between 4 and 5 kHz, and the constriction of consonants occurs at 7.5 kHz [12][13][14].…”

Section: Introductionmentioning

confidence: 99%

Semi-fragile digital speech watermarking for online speaker recognition

Nematollahi

Akhaee

Al-Haddad

et al. 2015

J AUDIO SPEECH MUSIC PROC.

View full text Add to dashboard Cite

In this paper, a semi-fragile and blind digital speech watermarking technique for online speaker recognition systems based on the discrete wavelet packet transform (DWPT) and quantization index modulation (QIM) has been proposed that enables embedding of the watermark within an angle of the wavelet's sub-bands. To minimize the degradation effects of the watermark, these sub-bands were selected from frequency ranges where little speaker-specific information was available (500-3500 Hz and 6000-7000 Hz). Experimental results on the TIMIT, MIT, and MOBIO speech databases show that the degradation results for speaker verification and identification are 0.39 and 0.97 %, respectively, which are negligible. In addition, the proposed watermark technique can provide the appropriate fragility required for different signal processing operations.

show abstract

Localization and selection of speaker-specific information with statistical modeling

Cited by 46 publications

References 20 publications

Human and computer recognition of regional accents and ethnic groups from British English speech

Human and computer recognition of regional accents and ethnic groups from British English speech

A Comparative Study of Methods for Handheld Speaker Verification in Realistic Noisy Conditions

Semi-fragile digital speech watermarking for online speaker recognition

Contact Info

Product

Resources

About