An Overview of the SPRACH System for the Transcription of Broadcast News
Columbia Academic Commons (Columbia University)Published 1 January 1999Open access
Gary Cook, James Christie, Daniel P. W. Ellis, Eric Fosler‐Lussier, Yoshi Gotoh, Brian Kingsbury
Citations25
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
This paper describes the SPRACH system, a system based on the connectionist-HMM framework and uses both recurrent neural network and multi-layer perceptron acoustic models, and describes recent developments to CHRONOS, a time-first stack decoder.
Abstract
This paper describes the SPRACH system developed for the 1998 Hub-4E broadcast news evaluation. The system is based on the connectionist-HMM framework and uses both recurrent neural network and multi-layer perceptron acoustic models. We describe both a system designed for the primary transcription hub, and a system for the less-than 10 times real-time spoke. We then describe recent developments to CHRONOS, a time-first stack decoder. We show how these developments have simplified the evaluation system, and led to significant reductions in the error rate of the 10x real-time system. We also present a system designed to operate in real-time with negligible search error.
Keywords
Computer Science
A post-processing system to yield reduced word error rates: Recognizer Output Voting Error Reduction (ROVER)
1,093 Citations2002J. Fiscus
A post-recognition process which models the output generated by multiple ASR systems as independent knowledge sources that can be combined and used to generate an output with reduced error rate.
Automatic Segmentation, Classification and Clustering of Broadcast News Audio
313 Citations1997M.A. Siegler
This work describes the problems faced in adapting a system built to recognize one utterance at a time to a task that requires recognition of an entire half hour show, and shows that a priori knowledge of acoustic conditions and speakers in the broadcast data is not required for segmentation.
Speech CommunicationRobust speech recognition using the modulation spectrogram
270 Citations1998Brian Kingsbury, Nelson Morgan +1 more
Using the modulation spectrogram as a front end for ASR provides a significant improvement in performance on highly reverberant speech and when it is used in combination with log-RASTA-PLP performance over a range of noisy and reverberant conditions is significantly improved, suggesting that the use of multiple representations is another promising method for improving the robustness of ASR systems.
Segment generation and clustering in the HTK broadcast news transcription system
92 Citations1998Thomas Hain, SE Johnson +3 more
This paper describes the segmentation, gender detection and segment clustering scheme used in the 1997 HTK broadcast news evaluation system and presents results on both the unpartitioned 1996 development and the 1997 evaluation sets.
Scalable backoff language models
86 Citations2002Kristie Seymore, Roni Rosenfeld
This project investigates the degradation of a trigram backoff model's perplexity and word error rates as bigram and trigram cutoffs are increased and shows that excluding trigrams and bigrams based on a weighted version of this difference method results in better perplexityand word error rate performance than excluding trigram and bigram based on counts alone.
Optimizing recognition and rejection performance in wordspotting systems
86 Citations2002H. Bourlard, Bart D'hoore +1 more
Compares the performance which can be achieved by different hidden Markov model (HMM) based wordspotting techniques when their parameters are tuned to optimize recognition and rejection rates and it appears that the proposed version performs at least as well as the other methods.
IEEE Transactions on Speech and Audio ProcessingStart-synchronous search for large vocabulary continuous speech recognition
31 Citations1999Steve Renals, M.M. Hochberg
A novel, efficient search strategy for large vocabulary continuous speech recognition, based on a stack decoder framework, that utilizes phone-level posterior probability estimates as a basis for phone deactivation pruning-a highly efficient method of reducing the required computation.
Time-first search for large vocabulary speech recognition
25 Citations2002T. Robinson, James Christie
A new search technique for large vocabulary speech recognition based on a stack decoder that proceeds time-first, that is partial path hypotheses are extended into the future in the inner loop and a tree walk over the lexicon is performed as an outer loop.
Confidence measures derived from an acceptor HMM
9 Citations1998Gethin Williams, Steve Renals
The results indicate that language model dependent confidence measures have reduced performance on BN data relative to that for the more grammatically constrained NAB data.
Reducing errors by increasing the error rate: MLP Acoustic Modeling for Broadcast News Transcription
8 Citations1999Nelson Morgan, Daniel P. W. Ellis +3 more
Some aspects of a Broadcast News recognition system based on hybrid HMM/MLP acoustic modeling are described, including the use of novel ‘modulation spectrogram’ features which are combined with conventional models at the posterior probability level, and an investigation of the interaction of model size and training set size for an multilayer perceptron (MLP) acoustic classifier.