Speech and language processing for next-millennium communications services

Cox, R.V.; Kamm, Candace; Rabiner, L. R.; Schroeter, Juergen; Wilpon, J. G.

doi:10.1109/5.880086

Cited by 47 publications

(21 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DTW is a pattern matching algorithm with a non-linear time optimization effect based on Bellman"s principle of optimality [19], which states that given an optimal path from A to B and a point C lying somewhere along this path, the path segments AC and CB are optimal paths from A to C and C to B respectively. The DTW problem can be formulated as follows: Given two time series X and Y of lengths |X| and |Y|, 1 2, …”

Section: Fig 6: Dtw Non-linear Alignment Of Two Time Seriesmentioning

confidence: 99%

See 1 more Smart Citation

Spoken Digits Recognition using Weighted MFCC and Improved Features for Dynamic Time Warping

Chapaneri¹

2012

IJCA

View full text Add to dashboard Cite

In this paper, we propose novel techniques for feature parameter extraction based on MFCC and feature recognition using dynamic time warping algorithm for application in speaker-independent isolated digits recognition. Using the proposed Weighted MFCC (WMFCC), we achieve low computational overhead for the feature recognition stage since we use only 13 weighted MFCC coefficients instead of the conventional 39 MFCC coefficients including the delta and double delta features. In order to capture the trends or patterns that a feature sequence presents during the alignment process, we compute the local and global features using Improved Features for DTW algorithm (IFDTW), rather than using the pure feature values or their estimated derivatives. The experiments based on TI-Digits corpus demonstrate the effectiveness of proposed techniques leading to higher recognition accuracy of 98.13%.

show abstract

Section: Fig 6: Dtw Non-linear Alignment Of Two Time Seriesmentioning

confidence: 99%

“…the speech signal, and classify it as a sequence of previously learned patterns, e.g. words or sub-word units such as phonemes [1]. Speech recognition systems can be characterized by many parameters, such as speaking model, speaking style, vocabulary, etc.…”

Section: Introductionmentioning

confidence: 99%

Spoken Digits Recognition using Weighted MFCC and Improved Features for Dynamic Time Warping

Chapaneri¹

2012

IJCA

View full text Add to dashboard Cite

show abstract

“…Real-time natural language translation, achieved by portable devices, is a major objective for the engineering community. Many years of research in this domain [2], [28], [45] have shown that the problem has solutions, but the required computational effort is high. This objective represents an important driver for multiprocessor architecture and related technologies.…”

Section: A Social and Economic Pullmentioning

confidence: 99%

An Outlook on Design Technologies for Future Integrated Systems

Micheli

2009

IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst.

View full text Add to dashboard Cite

Abstract-The economic and social demand for ubiquitous and multifaceted electronic systems-in combination with the unprecedented opportunities provided by the integration of various manufacturing technologies-is paving the way to a new class of heterogeneous integrated systems, with increased performance and connectedness and providing us with gateways to the living world. This paper surveys design requirements and solutions for heterogeneous systems and addresses design technologies for realizing them.

show abstract

“…After training during its manufacture, it would work "out of the box" to successfully recognize the utterances of different individuals speaking through a variety of channels. However, despite the steady progress of speech recognition technology in recent years, existing systems with large vocabularies are still sensitive to the nature of the acoustic environment and to the identity of the speaker [1][2][3] . For example, extensive retraining is often required if the acoustic channel is altered because the noise level changes, the speaker's room or position changes, or the signal conduit changes (telephone vs. room speech).…”

Section: Introductionmentioning

confidence: 99%

Blind normalization of speech from different channels

Levin

2004

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

We show how to construct a channel-independent representation of speech that has propagated through a noisy reverberant channel. This is done by blindly rescaling the cepstral time series by a non-linear function, with the form of this scale function being determined by previously encountered cepstra from that channel. The rescaled form of the time series is an invariant property of it in the following sense: it is unaffected if the time series is transformed by any time-independent invertible distortion.Because a linear channel with stationary noise and impulse response transforms cepstra in this way, the new technique can be used to remove the channel dependence of a cepstral time series. In experiments, the method achieved greater channelindependence than cepstral mean normalization, and it was comparable to the combination of cepstral mean normalization and spectral subtraction, despite the fact that no measurements of channel noise or reverberations were required (unlike spectral subtraction).

show abstract

Speech and language processing for next-millennium communications services

Cited by 47 publications

References 37 publications

Spoken Digits Recognition using Weighted MFCC and Improved Features for Dynamic Time Warping

Spoken Digits Recognition using Weighted MFCC and Improved Features for Dynamic Time Warping

An Outlook on Design Technologies for Future Integrated Systems

Blind normalization of speech from different channels

Contact Info

Product

Resources

About