New Techniques for a Real-time Phase Vocoder
Skip other details (including permanent urls, DOI, citation information)This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. Please contact mpub-help@umich.edu to use this work in a way not covered by the license. :
For more information, read Michigan Publishing's access and usage policy.
Page 178 ï~~NEW TECHNIQUES FOR REAL-T7 Ian Bowler Peter Manning DURHAM MUSIC TECHNOLOGY* Department of Music University of Durham Palace Green Durham DH1 3RL ENGLAND ABSTRACT: Some new techni Alan Pun Nick Baile DURHA School of University South Ro Durham I ENGLAN iques that are be real-time phase vocoder are described. r system to be implemented on a general-purpose R the analysis' part the vocoder, phase exploited to determine the frequency of spectral computationally efficient manner. Unlike some ot 17a 11TA. 11/ A a11 C a A1 T. a 1 S c./TAn x11 A l--.
Page 179 ï~~ANALYST A sound is sampled using an Analogue-to-Digit every R samples the L most recent samples are us being frame k As in the conventional phase multiplied by a window function to improve the f folded onto themselves B -1 times to produce a sui they are circularly rotated in order that the phas be relative to a common origin. The Fourier Trai into amplitude-phase form, the negative frequenc3 valued input signal. Each frame k thus produces and phase, Pk, where i indexes the frequency. The following procedure, which resolves the spectrum i a residual. Candidate sinusoids for frame k are selected by ct an amplitude peak:
Page 180 ï~~SYNTHESIS During synthesis, that part of the signal that is repr a set of oscillators forming an additive synthesiser.(B according to the following scheme: If a new sinusoida] allocated to an oscillator and its amplitude is increase the frame to measured amplitude at the end. If an exi through frame k, its original oscillator assignment disappears in frame k, its amplitude is reduced line4 beginning of the frame to zero in the middle. The rema from the residual by performing an inverse Fourier tra adding the result to the output of the additive synth( IMPLEMENTA2 The scheme is being implemented on a network of tran time operation at sample rates of up to 32kHz; futu will allow a 48kHz sample rate.