Research On Adjustment Of Speech Velocity, Volume And Tone In Chinese Speech Recognition

Human ear has the better capacity of self-adaptation, whose capacity of self-adaptation to the speech velocity, volume and tone is high. The speech recognition's system (in this paper we mainly discuss IBM ViaVoice) has the certain capacity of self-adaptation to the speech velocity, volume and tone, but the capacity of those is not enough with different enunciator. Too fast or too slow of the speech yelocity, too big or too small of volume and too high or too low of tone can reduce the rate of speech recognition. In order to deal with this problem, this paper introduces the author's research on some techniques related to speech processing, mainly including three aspects as follows:[1] In Chinese pronunciation, each syllable contains the vowel, the vowel's length is the main part in the syllable but the vowel doesn't contain the important information. According to these characteristics, we propose a method of adjusting the speech velocity by using similar waveform that is found by correlative coefficient in vowel part to lengthen or reduce the vowel part.[2] We adjust the volume by changing the swing of the sampling point according to average swing of the non-silence part.[3] The vowel's self-correlation function has periodicity and the period of this function is the fundamental sound period of this speech. According to this, we propose a method of adjusting the tone in temporal field by adding or deleting the sampling points in the waveform with the whole speech waveform unchanged.