FIELD: physics.
SUBSTANCE: acoustic signal undergoes wavelet transformation with calculation of wavelet coefficients which are then broken into segments of fixed duration. Fast Fourier transformation is used to find the Fourier spectrum of each segment of the wavelet coefficients and its energy is calculated and the boundary between vowel and consonant phonemes of the speech signal is determined. The most probable word candidates for the analysed signal are selected by comparing phonemes of the signal with phonemes of the word in the vocabulary.
EFFECT: possibility of recognising key words in a stream of continuous speech and faster operation of the system.
4 dwg
| Title | Year | Author | Number |
|---|---|---|---|
| METHOD OF SPEAKER-INDEPENDENT RECOGNITION OF KEY WORDS IN SPEECH | 2005 |
|
RU2294024C2 |
| SPEECH RECOGNITION METHOD BASED ON TWO-LEVEL MORPHOPHONEMIC PREFIX GRAPH | 2015 |
|
RU2597498C1 |
| SYSTEM AND METHOD OF SPEECH RECOGNITION | 2011 |
|
RU2466468C1 |
| METHOD OF RECOGNITION OF SEPARATE WORDS OF SPEECH WITH ADAPTATION TO ANNOUNCER | 1994 |
|
RU2047912C1 |
| METHOD FOR SPEAKER-INDEPENDENT RECOGNITION OF ISOLATED VOICE COMMANDS | 1997 |
|
RU2103753C1 |
| BIOMETRIC METHOD FOR SUBSCRIBER IDENTIFICATION BASED ON SPEECH SIGNAL | 2020 |
|
RU2742040C1 |
| METHOD FOR RECOGNIZING SPOKEN WORDS | 2005 |
|
RU2296376C2 |
| METHOD AND DEVICE FOR CLASSIFYING NOISY VOICE SEGMENTS USING MULTISPECTRAL ANALYSIS | 2014 |
|
RU2606566C2 |
| METHOD FOR HYBRID GENERATIVE-DISCRIMINATIVE SEGMENTATION OF SPEAKERS IN AUDIO-FLOW | 2013 |
|
RU2530314C1 |
| COMPUTER EQUIPMENT FOR READING OF PRINTED TEXT | 1996 |
|
RU2113726C1 |
