Speech Recognition by Machine

Acoust AINSWORTH architecture auditory automatic speech recognition Boltzmann machine branching factor cepstrum codebook continuous speech database demisyllables developed digital signal digital signal processor duration dynamic programming dynamic programming algorithm employed endpoint error rate estimated feature vectors filter bank formant frequencies frame fricatives function fundamental frequency hidden Markov models hypothesiser IEEE IEEE ICASSP IEEE Trans input isolated word isolated-word knowledge sources lexical linear prediction analysis linear prediction coefficients machine Markov models matching microphone noise normalisation obtained operation parameters pattern peaks perception performance phoneme plosives possible problem Proc processors produced Rabiner recog recognition accuracy recognition rate Sakoe second formant Section segment semantic sentence sequence shown in Fig signal processing speaker-independent speakers spectral spectrum speech signal speech synthesis spoken syllables syntactic syntax synthesiser systolic array task techniques templates test utterance test word tion vocabulary vocal tract voice voiceless vowels word recognition