250 likes | 384 Views
Analysis & Synthesis. The Vocoder and its related technology. (Extremely) Simplified Model of Speech Production. Periodic source. voiced. Filters. Coupling. Speech. unvoiced. Noise source. (Slightly less) Simplified Model of Speech Production. pitch. Impulse train. V/uv bit.
E N D
Analysis & Synthesis The Vocoder and its related technology
(Extremely) Simplified Model of Speech Production Periodic source voiced Filters Coupling Speech unvoiced Noise source
(Slightly less) Simplified Model of Speech Production pitch Impulse train V/uv bit Glottal model Filter spec voiced Gain v Coupling x Filters Speech Gain u x unvoiced Noise source
Glottal Waveform Caricature Glottis closed (decoupled) Glottis open (nonlinear)
Vocoders: vocal tract synthesis • Short time spectral synthesis • Linear predictive synthesis • Cepstral synthesis • Formant synthesis
Vocoders: vocal tract analysis • Short time spectral analysis • Linear predictive analysis • Cepstral analysis • Formant analysis
Vocoders: vocal source analysis • Pitch detection • Voiced/unvoiced decision
Figure 3.5: Narrow band spectrogram Narrowband spectrogram
Formant Analysis • Peak picking from smoothed spectrum • Root-finding from LPC polynomial - finding second-order sections • In general, position more important than bandwidth • Deterministic or statistical analysis to find best formant “track”
Ed Lee Vocoder Page http://ptolemy.eecs.berkeley.edu/~eal/audio/vocoder.html
Linux Vocoder Page http://www.sirlab.de/linux/descr_vocoder.html
How does the speech sound with the middle spectrum differ from the left one? From the right?