Low-band extension of telephone-band speech

Miet, G.; Gerrits, A.J.; Valière, Jean-Christophe

doi:10.1109/icassp.2000.862116

Cited by 23 publications

(14 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Many studies are motivated by telephone companies interested in the reconstruction of wideband speech (with frequencies ranging 0.05-7kHz) from a limited bandwidth transmission (0.3-3.4kHz). Low-band and highband signal representations are inferred from the transmitted region, assuming a high degree of correlation, and are added to the input signal to reconstruct a wideband signal [1,2]. Similar studies exist for in-vehicle communications and other noisy environments [3,4].…”

Section: Introductionmentioning

confidence: 99%

MFCC Compensation for Improved Recognition of Filtered and Band-Limited Speech

Morales

Hansen

Toledano³

Proceedings. (ICASSP '05). IEEE International Conference on Acoustics, Speech, and Signal Processing, 2005.

View full text Add to dashboard Cite

This paper addresses the problem of bandwidth expansion for the purpose of robust speech recognition. We show that an HMM-based ASR engine trained with full spectrum range data (0-8kHz) can successfully perform speech recognition tasks over band-filtered test data compensated by means of a series of simple MFCC parameter corrector functions. The problem is important when ASR is employed for audio streams of unknown frequency bandwidth common in spoken document retrieval. Evaluation is based on recognition rates. Accuracy varies depending on the width and spectral regions eliminated, but the system shows great advantages over the use of uncompensated filtered test data. The theoretical maximum recognition rates using corrector functions over filtered test data are very close to the base rate (unfiltered data) even when the greatest part of the spectrum of the original data is suppressed. These rates are even better than those obtained in the matched train/test HMMs with filtered data.

show abstract

Section: Introductionmentioning

confidence: 99%

MFCC Compensation for Improved Recognition of Filtered and Band-Limited Speech

Morales

Hansen

Toledano³

Proceedings. (ICASSP '05). IEEE International Conference on Acoustics, Speech, and Signal Processing, 2005.

View full text Add to dashboard Cite

show abstract

“…Miet et al [6] described a technique which splits the telephoneband speech signal into a spectral envelope and a short-term residual. Where the spectral envelope and the residual are extended separately and recombined to create an extended band signal.…”

Section: Introductionmentioning

confidence: 99%

Simulation of Narrow Band Speech Signal using BPN Networks

Gandhimathi¹,

Narmadha²,

Lakshmi³

2010

IJCA

View full text Add to dashboard Cite

This paper proposes to extend the band width of narrow band telephone speech signal by employing feed forward back propagation neural network. There are different types of faster training algorithm are available in the literature like Variable Learning Rate, Resilient Back propagation, Polak-Ribiére Conjugate Gradient , Conjugate Gradient with Powell/Beale Restarts , BFGS Quasi-Newton , One-Step Secant , FletcherPowell Conjugate Gradient Algorithms, Scaled Conjugate Gradient and Liebenberg-Marquardt. These algorithms are used to train the BPN networks using Neural network tool box. The correlation between the inputs of the neural network and the input-output correlation were calculated. The components were employed to reconstruct the speech signal and the results are analyzed. General TermsArtificial band width expansion, Signal processing, Neural networks , Training algorithm . KeywordsAR Filter, Back propagation neural network , linear mapping method, code book method.

show abstract

“…Proposed solutions for highband-spectral envelope extension range from a simple fixed envelope method [2,3], to linear mapping [7], to more complex methods such as Neural Networks [6,11], Statistical Recovery [12], Hidden Markov Model (HMM) [13] as well as VQ [4,8,9,14].…”

Section: Envelope Extension Using Vqmentioning

confidence: 99%

“…In practice, Linear Predictive Coefficients, Line Spectral Pairs [14], Log Envelope Samples [9] or cepstral coefficients [4,7] are used instead. In this paper, we map 8 narrowband cepstral coefficients into 16 wideband cepstral coefficients.…”

Section: Envelope Extension Using Vqmentioning

confidence: 99%

Bandwidth Extension of Narrowband Speech using Soft-decision Vector Quantization

Soon

Yeo

2005 5th International Conference on Information Communications &Amp; Signal Processing

View full text Add to dashboard Cite

Vector Quantization (VQ) is normally used to extend the spectral envelope of narrowband speech into that of the wideband speech. This paper proposes a novel soft-decision approach for VQ and it can be shown that the result is better than the direct method as well as the interpolated VQ method. The resulting cepstral wideband coefficients are obtained using a weighted combination of several set of coefficients in the codebook. The weighting factors are determined by an inversedistance formula.

show abstract

Low-band extension of telephone-band speech

Cited by 23 publications

References 3 publications

MFCC Compensation for Improved Recognition of Filtered and Band-Limited Speech

MFCC Compensation for Improved Recognition of Filtered and Band-Limited Speech

Simulation of Narrow Band Speech Signal using BPN Networks

Bandwidth Extension of Narrowband Speech using Soft-decision Vector Quantization

Contact Info

Product

Resources

About