Publications

SUMMARY: Enchancement of noisy speech could be optimally performed if the probability distribution's (PD's) of the source and the noise were known.

Two methods for protection of tone signalling against interference from noise of all descriptions within the frequency band received are feasible.

I N the past, much research has been devoted to the determination of the relative magnitudes of the frequency components of speech, and the results of these explorations are useful and well known.

The paper is chiefly concerned with the important speech power quantities--frequency spectra, distributions of instantaneous, average, syllabic and peak amplitudes, etc.--as they obtain in actual s

At AT&T Bell Laboratories, we have studied a broad range of systems for speech recognition.

Algorithms for speech recognition can be characterized broadly as pattern recognition approaches and acoustic phonetic approaches.

Future wireless multimedia terminals will have a variety of applications that require speech recognition capabilities.

We describe a large-vocabulary speaker-independent continuous speech recognition system being implemented on MARS, a multiprocessor developed at AT&T Bell Laboratories [1,2].

The development of robust, speaker-independent, automatic speech recognition (ASR) systems that perform well over dialed-up telephone lines has been a topic of interest at AT&T for over a decad

At AT&T Bell Laboratories, we have studied a broad range of systems for speech recognition.