Publications

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

At AT&T Bell Laboratories, we have studied a broad range of systems for speech recognition, depending on the intended application area.

Multi-layered neural networks have been recently proposed for non-linear signal processing, in particular, non-linear prediction and system modeling [1].

The characterization of speech signal power on Bell System message circuits is an essential step in the determination of signal power loading and crosstalk objectives.

Speech synthesis by rule is the method of converting from a discrete representation of speech in linguistic units, that is, phonemes and stress marks, to a continuous acoustic waveform. Fig.

Algorithms utilizing complementary strengths of waveform coding and vocoding principles have demonstrated the possibility of speech communication with high quality at rates below 16 kb/s.

T h e volume of message signals at various points in the telephone network is of importance to those who design and engineer telephone systems and equipment, and ultimately, of course, to the liste

Describes in detail with studio plan and block switching diagrams the layout of the building, method of programme switching and signalling circuits.

A new algorithmic technique is presented for efficiently implementing the end-point decisions necessary to separate and segment speech from noisy background environments.

The effectiveness of surface recombination in speeding up the relaxation of a GaAs etalon is reported.