Phase Vocoder. Flanagan, J., L.; Meinhart, D., I., S.; Golden, R., M.; and Sondhi, M., M. The Journal of the Acoustical Society of America, 38(5):939-940, ASA, 11, 1965.
Phase Vocoder [pdf]Paper  abstract   bibtex   
Atype of vocoder is described that promises modest bandsaving andelimination of the pitch-tracking and voiced-unvoiced switching inherent in spectrumchannel vocoders. A speech signal f(t) suffers little degradation whenpassed through a parallel bank of contiguous bandpass filters andthen recombined. If fn(t) is the output of the nthbandpass filter, the original signal f(t) is approximated by nfn(t).Each fn(t) can be represented by two parameters: the valueof the short-time amplitude spectrum of f(t) evaluated at frequencyn, and the time derivative of the short-time phase spectrumalso evaluated at n. These data are transmitted to asynthesizer that produces approximations to each fn(t). The complex short-timespectrum calculated for each channel is , where h(t) isthe impulse response of a realizable low-pass filter. The amplitude|F(n,t)| and the phase derivative (n,t) are formed for eachchannel, low-pass-filtered and transmitted to the sythesizer. At the synthesizer,(n,t) frequency-modulates an oscillator of nominal center frequency n, and|F(n,t)| amplitude-modulates the same oscillator. The synthesized signals for alln channels are then summed. In effect, the |F(n,t)| signalscarry the spectral envelope information, and the (n,t) signals carrythe excitation information. Previous experience with channel vocoders shows thatthe |F(n,t)| signals may be band-limited to around 20 cps.Our experiments with the phase vocoder indicate that the (n,t)signals may be similarly band-limited. Speech transmitted by the phasevocoder is demonstrated. ©1965 Acoustical Society of America
@article{
 id = {d3316f72-4c5e-3682-868c-b58c87edc556},
 title = {Phase Vocoder},
 type = {article},
 year = {1965},
 created = {2010-05-05T19:31:22.000Z},
 pages = {939-940},
 volume = {38},
 websites = {http://link.aip.org/link/?JAS/38/939/4},
 month = {11},
 publisher = {ASA},
 file_attached = {true},
 profile_id = {6d353feb-efe4-367e-84a2-0815eb9ca878},
 group_id = {7f22d395-e966-3e55-9f90-9d0061e6682a},
 last_modified = {2012-08-03T21:33:23.000Z},
 read = {true},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Flanagan1965},
 short_title = {J. Acoust. Soc. Am.},
 client_data = {"desktop_id":"3608ffc0-efd0-4adf-8523-3306b8b1add5"},
 abstract = {Atype of vocoder is described that promises modest bandsaving andelimination of the pitch-tracking and voiced-unvoiced switching inherent in spectrumchannel vocoders. A speech signal f(t) suffers little degradation whenpassed through a parallel bank of contiguous bandpass filters andthen recombined. If fn(t) is the output of the nthbandpass filter, the original signal f(t) is approximated by nfn(t).Each fn(t) can be represented by two parameters: the valueof the short-time amplitude spectrum of f(t) evaluated at frequencyn, and the time derivative of the short-time phase spectrumalso evaluated at n. These data are transmitted to asynthesizer that produces approximations to each fn(t). The complex short-timespectrum calculated for each channel is , where h(t) isthe impulse response of a realizable low-pass filter. The amplitude|F(n,t)| and the phase derivative (n,t) are formed for eachchannel, low-pass-filtered and transmitted to the sythesizer. At the synthesizer,(n,t) frequency-modulates an oscillator of nominal center frequency n, and|F(n,t)| amplitude-modulates the same oscillator. The synthesized signals for alln channels are then summed. In effect, the |F(n,t)| signalscarry the spectral envelope information, and the (n,t) signals carrythe excitation information. Previous experience with channel vocoders shows thatthe |F(n,t)| signals may be band-limited to around 20 cps.Our experiments with the phase vocoder indicate that the (n,t)signals may be similarly band-limited. Speech transmitted by the phasevocoder is demonstrated.

        
         ©1965 Acoustical Society of America},
 bibtype = {article},
 author = {Flanagan, J. L. and Meinhart, D. I. S. and Golden, Roger M. and Sondhi, Man Mohan},
 journal = {The Journal of the Acoustical Society of America},
 number = {5}
}
Downloads: 0