SPEECH-RECOGNITION CIRCUITRY EMPLOYING NONLINEAR PROCESSING, SPEECH ELEMENT MODELING AND PHONEME ESTIMATION

摘要

A phoneme estimator in a speech-recognition systemincludes energy detect circuitry for detecting the segments ofa speech signal that should be analyzed for phoneme content.Speech-element processors then process the speech signalsegments, calculating nonlinear representations of thesegments. The nonlinear representation data is applied tospeech-element modeling circuitry which reduces the datathrough speech element specific modeling. The reduced data arethen subjected to further nonlinear processing. The results ofthe further nonlinear processing are again applied to speech-element modeling circuitry, producing phoneme isotypeestimates. The phoneme isotype estimates are rearranged andconsolidated, that is, the estimates are uniformly labeled andduplicated estimates are consolidated, forming estimates ofwords or phrases containing minimal numbers of phonemes. Theestimates may then be compared with stored words or phrases todetermine what was spoken.