Lemur Analysis Frame Length
During Lemur analysis, a series of short-time spectra of windowed
samples is computed using a fast Fourier transform (FFT), and the magnitude and phase are computed from the
complex FFT data (Oppenheim and Schafer 1989).
Each spectrum corresponds to an analysis frame, a list of the significant spectral
components in the analysis signal at the time of that spectrum. The Frame Length is the separation (in time)
between successive spectra. Generally, the analysis windows overlap considerably, so the frame length is much
less than the window length. During synthesis, consecutive analysis frames are used to generate a frame of
samples, that is, a number of samples corresponding to the frame length. Two frames of analysis data are needed
to synthesize one frame of samples. The frame length is, thus, a lower bound on the rate of change of the spectral
character of sounds synthesized in Lemur. Lemur track parameters modulate smoothly between the values
measured at times separated by the frame length. Thus, for signals with rapidly changing spectra, short frames
are needed to preserve the temporal information at synthesis time, whereas signals that are relatively stationary
may be analyzed with longer analysis frames.
The Frame Length for a Lemur analysis may be set directly in the
Spectral Analysis Parameters panel on Lemur's Parameters menu.
Forward to Lemur Analysis Thresholding
Back to Windowing In Lemur Anlaysis
Up to Lemur
Documentation Index
lemur@uiuc.edu