Mitchel Weintraub - Academia.edu (original) (raw)

Uploads

Papers by Mitchel Weintraub

IEEE International Conference on Acoustics Speech and Signal Processing, 1993

... 6. 7. 8. 9. REFERENCES H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI&#... more ... 6. 7. 8. 9. REFERENCES H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI'sDECIPHER Speech Recognition System on DAR-PA's CSR Task,” 1992 DARPA Speech and Natural Lan-guage Workshop Proceedings, pp 410-414 Murveit, H., J. Butzberger, and M ...

The Psychophysics of Speech Perception, 1987

... lnstead, the two people will take turns talking to each other. When listening to this type of... more ... lnstead, the two people will take turns talking to each other. When listening to this type of conversation, the original goal of a sound separation system (computing a ... The computational model for separating sounds is based on the following principle: The information computed by ...

ICASSP '86. IEEE International Conference on Acoustics, Speech, and Signal Processing, 1986

This paper describes a computational model that attempts to separate two simultaneous talkers. Th... more This paper describes a computational model that attempts to separate two simultaneous talkers. The goal of this model is to improve a speech recognition system's ability to recognize what each of the two talkers say. The model consists of the following stages: (1) an iterative dynamic programming algorithm to track the pitch period for each of the two talkers, (2) a Markov model to determine the characteristics (e.g. voiced-unvoiced) of each speaker's voice, (3) a recursive algorithm that uses both local periodicity information and local spectral continuity constraints to compute a spectral estimate of each talker, (4) a resynthesis algorithm to convert the spectral estimate of each talker into a speech waveform, and (5) a speaker-independent continuous-digit-recognition system that attempts to recognize what each of two talkers is saying. The system was trained and tested on a database of simultaneous digit strings spoken by a male and female talker. An evaluation of the different stages of this model is presented.

Proceedings of the workshop on Human Language Technology - HLT '93, 1993

1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, 1997

The paper describes the methods and results of a study of the feasibility of automatically gradin... more The paper describes the methods and results of a study of the feasibility of automatically grading the performance of Japanese students when reading English aloud. SRI recorded 31 adult Japanese speakers: 22 men and 9 women. Each Japanese speaker read six sentences aloud. All ...

An algorithm based on hidden Markov models is applied to the task of speaker-independent continuo... more An algorithm based on hidden Markov models is applied to the task of speaker-independent continuous-speech recognition for a vocabulary of 1000 words with no syntactic constraints. The signal is limited to 4000 Hz. Word models were built from three-state representations of phonetic units, concatenated according to entries in a lexicon. Performance as measured on DARPAs resource management database was 40% correct word recognition. It was found that the use of several different acoustic features and the use of word-specific phonetic modeling, where possible, improved system performance

LVCSR LOG-LIKELIHOOD RATIO SCORING FOR KEYWORD SPOTTING ... Our word recognition error rate on th... more LVCSR LOG-LIKELIHOOD RATIO SCORING FOR KEYWORD SPOTTING ... Our word recognition error rate on this same test set is 54.7%. ... Using the one best answer from the Viterbi backtrace, we used the average probability per frame as the score for each hypothesized keyword ...

... thesis and that in my opinion it is fully adequate, in scope and quality, as a dissertation f... more ... thesis and that in my opinion it is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ... Doree Weintraub, Zachary Weintraub, Leonard Weintraub, Mom and Dad for providing me with t~e love and emotional support to complete this thesis. ...

Proceedings of the workshop on Human Language Technology - HLT '94, 1994

Proceedings of International Conference on Neural Networks (ICNN'97), 1997

We compare two methods for modeling context in the framework of a hybrid hidden Markov model (HMM... more We compare two methods for modeling context in the framework of a hybrid hidden Markov model (HMM)/multilayer perceptron (MLP) speaker-independent continuous speech recognition system. The first method for modeling context is based on the computation of HMM context-dependent observation probabilities using a Bayesian factorization in terms of scaled posterior phone probabilities that are computed with a set of MLPs, one for every relevant context. The second method is based on the use of input features composed of extended multiframe windows of acoustic vectors that include the acoustic information of the current phone as well as various degrees of the acoustic information of the adjacent left and right phones. Experimental results using a hybrid HMM-MLP speaker-independent continuous speech recognition system show that the first approach, based on connectionist context-dependent estimation of observation probabilities, is more efficient in the use of parameters for the same level of recognition performance

IEEE International Conference on Acoustics Speech and Signal Processing, 1993

... 727-730 H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI's D... more ... 727-730 H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI's DECIPHER Speech Recognition System on DARPA's CSR Task,” 1992 DARPA Speech and Natural Language Workshop Proceedings, pp 410-414 Murveit, H., J. Butzberger, and M. Weintraub. ...

International Conference on Acoustics, Speech, and Signal Processing, 1989

In this paper we show that the performance of hidden-Markov model (HMM) speech recognition system... more In this paper we show that the performance of hidden-Markov model (HMM) speech recognition systems can be improved through appropriate incorporation of speech and linguistic knowledge. The HMM formulation is a powerful statistical model that has been shown to ...

Proceedings of the workshop on Speech and Natural Language - HLT '90, 1990

SRI and U.C. Berkeley have begun a cooperative effort to develop a new architecture for real-time... more SRI and U.C. Berkeley have begun a cooperative effort to develop a new architecture for real-time implementation of spoken language systems (SLS). Our goal is to develop fast speech recognition algorithms, and supporting hardware capable of recognizing continuous speech from a bigram-or trigram-based 20,000-word vocabulary or a 1,000-to 5,000word SLS.

IEEE International Conference on Acoustics Speech and Signal Processing, 1993

... 6. 7. 8. 9. REFERENCES H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI&#... more ... 6. 7. 8. 9. REFERENCES H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI'sDECIPHER Speech Recognition System on DAR-PA's CSR Task,” 1992 DARPA Speech and Natural Lan-guage Workshop Proceedings, pp 410-414 Murveit, H., J. Butzberger, and M ...

The Psychophysics of Speech Perception, 1987

... lnstead, the two people will take turns talking to each other. When listening to this type of... more ... lnstead, the two people will take turns talking to each other. When listening to this type of conversation, the original goal of a sound separation system (computing a ... The computational model for separating sounds is based on the following principle: The information computed by ...

ICASSP '86. IEEE International Conference on Acoustics, Speech, and Signal Processing, 1986

This paper describes a computational model that attempts to separate two simultaneous talkers. Th... more This paper describes a computational model that attempts to separate two simultaneous talkers. The goal of this model is to improve a speech recognition system's ability to recognize what each of the two talkers say. The model consists of the following stages: (1) an iterative dynamic programming algorithm to track the pitch period for each of the two talkers, (2) a Markov model to determine the characteristics (e.g. voiced-unvoiced) of each speaker's voice, (3) a recursive algorithm that uses both local periodicity information and local spectral continuity constraints to compute a spectral estimate of each talker, (4) a resynthesis algorithm to convert the spectral estimate of each talker into a speech waveform, and (5) a speaker-independent continuous-digit-recognition system that attempts to recognize what each of two talkers is saying. The system was trained and tested on a database of simultaneous digit strings spoken by a male and female talker. An evaluation of the different stages of this model is presented.

Proceedings of the workshop on Human Language Technology - HLT '93, 1993

1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, 1997

The paper describes the methods and results of a study of the feasibility of automatically gradin... more The paper describes the methods and results of a study of the feasibility of automatically grading the performance of Japanese students when reading English aloud. SRI recorded 31 adult Japanese speakers: 22 men and 9 women. Each Japanese speaker read six sentences aloud. All ...

An algorithm based on hidden Markov models is applied to the task of speaker-independent continuo... more An algorithm based on hidden Markov models is applied to the task of speaker-independent continuous-speech recognition for a vocabulary of 1000 words with no syntactic constraints. The signal is limited to 4000 Hz. Word models were built from three-state representations of phonetic units, concatenated according to entries in a lexicon. Performance as measured on DARPAs resource management database was 40% correct word recognition. It was found that the use of several different acoustic features and the use of word-specific phonetic modeling, where possible, improved system performance

LVCSR LOG-LIKELIHOOD RATIO SCORING FOR KEYWORD SPOTTING ... Our word recognition error rate on th... more LVCSR LOG-LIKELIHOOD RATIO SCORING FOR KEYWORD SPOTTING ... Our word recognition error rate on this same test set is 54.7%. ... Using the one best answer from the Viterbi backtrace, we used the average probability per frame as the score for each hypothesized keyword ...

... thesis and that in my opinion it is fully adequate, in scope and quality, as a dissertation f... more ... thesis and that in my opinion it is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ... Doree Weintraub, Zachary Weintraub, Leonard Weintraub, Mom and Dad for providing me with t~e love and emotional support to complete this thesis. ...

Proceedings of the workshop on Human Language Technology - HLT '94, 1994

Proceedings of International Conference on Neural Networks (ICNN'97), 1997

We compare two methods for modeling context in the framework of a hybrid hidden Markov model (HMM... more We compare two methods for modeling context in the framework of a hybrid hidden Markov model (HMM)/multilayer perceptron (MLP) speaker-independent continuous speech recognition system. The first method for modeling context is based on the computation of HMM context-dependent observation probabilities using a Bayesian factorization in terms of scaled posterior phone probabilities that are computed with a set of MLPs, one for every relevant context. The second method is based on the use of input features composed of extended multiframe windows of acoustic vectors that include the acoustic information of the current phone as well as various degrees of the acoustic information of the adjacent left and right phones. Experimental results using a hybrid HMM-MLP speaker-independent continuous speech recognition system show that the first approach, based on connectionist context-dependent estimation of observation probabilities, is more efficient in the use of parameters for the same level of recognition performance

IEEE International Conference on Acoustics Speech and Signal Processing, 1993

... 727-730 H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI's D... more ... 727-730 H. Murveit, J. Butzberger, and M. Weintraub, “Performance of SRI's DECIPHER Speech Recognition System on DARPA's CSR Task,” 1992 DARPA Speech and Natural Language Workshop Proceedings, pp 410-414 Murveit, H., J. Butzberger, and M. Weintraub. ...

International Conference on Acoustics, Speech, and Signal Processing, 1989

In this paper we show that the performance of hidden-Markov model (HMM) speech recognition system... more In this paper we show that the performance of hidden-Markov model (HMM) speech recognition systems can be improved through appropriate incorporation of speech and linguistic knowledge. The HMM formulation is a powerful statistical model that has been shown to ...

Proceedings of the workshop on Speech and Natural Language - HLT '90, 1990

SRI and U.C. Berkeley have begun a cooperative effort to develop a new architecture for real-time... more SRI and U.C. Berkeley have begun a cooperative effort to develop a new architecture for real-time implementation of spoken language systems (SLS). Our goal is to develop fast speech recognition algorithms, and supporting hardware capable of recognizing continuous speech from a bigram-or trigram-based 20,000-word vocabulary or a 1,000-to 5,000word SLS.