Phonetics three branches:
1. phonetic pronunciation
2. The acoustic phonetics ---- further emergence of analog sound, speech synthesis, speech recognition
3. auditory phonetics and psycholinguistics ---- the human ear and brain research
This chapter describes the process of speech production and human hearing process, the traditional linear and non-linear speech production model widely attention of
Speech production models, which are engaged in basic research knowledge processing voice signals.
2.1 generating speech signal
Vocal cord vibration generating a sound, which is the basic sound source generates sound source called the vocal cords.
Pitch period
Pitch frequency 80Hz - 500Hz
Channel
Voiced sound voiced vocal cord vibration generated
Voiceless unvoiced sound
Speech is the way sound waves traveling through air. Is a longitudinal acoustic wave, its vibration direction and the propagation direction is the same .
Complex sound
Pure tone - only for a pitch, no overtone.
French physicist Fourier discovered the differences between each sound is polyphonic different (chord) of.
Was able to hear each instrument has its own special sound , it is because they have different chords.
A voice pitch and overtone, composed the sound of chords.
In the composite tone, a fundamental frequency of the lowest frequency, but the maximum amplitude. Each overtone remaining energy gradually reduced, the amplitude gradually decreases.
2.1.3 speech signal represented in the time domain and frequency domain
Spectrum yes yes basic parameters characterizing speech characteristics. Wherein the resonance peak is a typical frequency domain parameter, which can determine the signal spectrum
Overall profile or spectral envelope.
It is generally assumed that the speech signal is a short-time stationary signal . Time-frequency analysis, wavelet transform
Frequency spectrogram view ----
Chinese speech prosody characteristics 2.1.5
Speech acoustic feature refers to strong timbre, pitch, tone length and tone.
Chinese, mainly by timbre and pitch to distinguish semantics, sound intensity and sound length can not be distinguished semantics.