Speech recognition system

摘要: An acoustic signal processing circuit extracts input speech pattern data and subsidiary feature from an signal. The comprise frequency spectra, whereas the phoneme features. These are then stored in a buffer memory. similarity measures between memory reference dictionary computed by computation circuit. When largest measure exceeds first threshold value when difference second value, category of which gives is produced control to correspond speech. recognition cannot be performed, categories patterns respectively give mth compared with data. In this manner, voice performed section.