Showing papers by "John Makhoul published in 1992"

BBN BYBLOS and HARC February 1992 ATIS benchmark results

Abstract: We present results from the February '92 evaluation on the ATIS travel planning domain for HARC, the BBN spoken language system (SLS). In addition, we discuss in detail the individual performance of BYBLOS, the speech recognition (SPREC) component.In the official scoring, conducted by NIST, BBN's HARC system produced a weighted SLS score of 43.7 on all 687 evaluable utterances in the test set. This was the lowest error achieved by any of the 7 systems evaluated.For the SPREC evaluation BBN's BYBLOS system achieved a word error rate of 6.2% on the same 687 utterances and 9.4% on the entire test set of 971 utterances. These results were significantly better than any other speech system evaluated.

Continuous speech recognition using segmental neural nets

Abstract: The authors present the concept of a 'segmental neural net' (SNN) for phonetic modeling in continuous speech recognition (CSR) and show how this can be used, together with a hidden Markov model (HMM) system, to improve continuous speech recognition (CSR). The SNN is a segment-based model that uses a neural network to correlate features of the speech signal throughout the duration of a phonetic segment. The problem of handling phonetic segments of varying length is solved by applying a warping function which provides the neural network inputs with a fixed-length representation of the segment. This method of modeling speech differs from that of HMMs, which assume that speech frames are conditionally independent. To take advantage of the training and decoding speed of HMMs, a hybrid SNN/HMM system has been developed to combine the advantages of both types of approaches. In this hybrid system, use is made of the N-best paradigm to generate likely phonetic segmentations, which are then scored by the SNN. The HMM and SNN scores are then combined to optimize performance. >

Robust continuous speech recognition

Abstract: The primary objective of this basic research program is to develop robust methods and models for speaker-independent acoustic recognition of spontaneously-produced, continuous speech. The work has focussed on developing accurate and detailed models of phonemes and their coarticulation for the purpose of large-vocabulary continuous speech recognition. Important goals of this work are to achieve the highest possible word recognition accuracy in continuous speech and to develop methods for the rapid adaptation of phonetic models to the voice of a new speaker.

A Hybrid Neural Net System for State-of-the-Art Continuous Speech Recognition

Abstract: Untill recently, state-of-the-art, large-vocabulary, continuous speech recognition (CSR) has employed Hidden Markov Modeling (HMM) to model speech sounds. In an attempt to improve over HMM we developed a hybrid system that integrates HMM technology with neural networks. We present the concept of a "Segmental Neural Net" (SNN) for phonetic modeling in CSR. By taking into account all the frames of a phonetic segment simultaneously, the SNN overcomes the well-known conditional-independence limitation of HMMs. In several speaker-independent experiments with the DARPA Resource Management corpus, the hybrid system showed a consistent improvement in performance over the baseline HMM system.

Improving state-of-the-art continuous speech recognition systems using the N-best paradigm with neural networks

Abstract: In an effort to advance the state of the art in continuous speech recognition employing hidden Markov models (HMM), Segmental Neural Nets (SNN) were introduced recently to ameliorate the well-known limitations of HMMs, namely, the conditional-independence limitation and the relative difficulty with which HMMs can handle segmental features. We describe a hybrid SNN/HMM system that combines the speed and performance of our HMM system with the segmental modeling capabilities of SNNs. The integration of the two acoustic modeling techniques is achieved successfully via the N-best rescoring paradigm. The N-best lists are used not only for recognition, but also during training. This discriminative training using N-best is demonstrated to improve performance. When tested on the DARPA Resource Management speaker-independent corpus, the hybrid SNN/HMM system decreases the error by about 20% compared to the state-of-the-art HMM system.

BBN real-time speech recognition demonstrations

Abstract: Typically, real-time speech recognition -- if achieved at all -- is accomplished either by greatly simplifying the processing to be done, or by the use of special-purpose hardware. Each of these approaches has obvious problems. The former results in a substantial loss in accuracy, while the latter often results in obsolete hardware being developed at great expense and delay.