The partial tracking analysis offers two methods, thtat apply to harmonic or inharmonic sounds. In addition to the general FFT settings that characterize all the analysis offered in AS, a number of parameters are common, or specific to both methods.
This analysis can be used prior to a chord sequence analysis. It is a prerequisite to partial synthesis.
In the In the upper pop up menu, choose the desired item. |
Maximum number of partials per analysis window to detect.
Minimum amplitude in dB required to detect a partial. You can use the slider, but you can also use the white threshold of the sonogram display.
Preview option
: when moving the slider, the sonogram window contrast will be adjusted just as if using the white threshold slider of the sonogram display.
Link to sonogram
: the threshold is set according to the current contrast of the sonogram, as defined with the white threshold slider. Note that the contrast cannot be modified while setting the analysis parameters in the dialogue window. It must be done before.
This enveloppe adds a fade in or fade out before and after the partial.
Smoothing Envelope Attack : enveloppe at the beginning of the partial
Smoothing Envelope Release : enveloppe at the end of the partial
By default, this parameter is not selected.
Chord sequence analysis : the analysis is related to the partial tracking, but additionnally, it allows to gather partials into coherent groups, like chords. Consequentyl, the smoothing envelope parameter is not relevant for this type of analysis, and is not provided. If the partial tracking analysis is used as a means to study the spectral contents before executing a chord sequence analysis, the smoothing envelope shouldn't be used.
Partial synthesis : once an enveloppe is defined during the analysis, it will affect the result permanently. Other smoothing envelope parameters can be defined for the partials synthesis, but the impact of the envelope selected for the tracking analysis can't be changed. If you plan to synthesize the partials, It might be wiser not to use any envelope for the analysis, and to apply an envelope for the synthesis only, as a more flexible solution.
The approximation of the true spectral peak shape is more or less biased, affecting the estimation of the peak position, frequency or amplitude for instance. This can be corrected with several algorithms, using the quadratic interpolation and zero padding.
Three methods can be chosen via the Bias Correction
pop up menu :
Simple Demodulation
Approximate
Advanced Demodulation
These three methods are respectively gradually more efficient, but are also more costly in terms of calculus.
Even if the calculus is interrupted, the part of the analysis that has already been executed will be displayed.
In the upper pop up menu, choose Inharmonic Analysis. This method is selected by default.
The parameters of the peak connection thresholds are related with the maximum likelihood... They determine the maximum tolerated frequency and amplitude variations to connect two peaks from an analysis window to another.
Frequency Deviation : maximum range between two peaks. The deviation can be expressed as a relative difference, in cents, or as a constant difference in Hz. By default, this value is equal to 20 cents – 1/10 tone – or 0 Hz.
Relative Amplitude Deviation : maximum relative percentage of amplitude between two peaks. The default value is set to 50 %.
Source and Target Partial Neighbours : number of possible connectable neighbour peaks, with a source, and a target. By default, one source and three target peaks can be connected to each other.
This parameter applies to fragments of sinusoids. The parameters are quite alike the peak connection parameters.
Prior to performing the partial tracking, a fundamental frequency (F0) analysis must be done.
In the Analysis menu, choose Fundamental Frequency.
Execute the analysis – see corresponding section –.
Save the F0 as an sdif file.
In the upper pop up menu, choose Harmonic Analysis
.
In the Fundamental Analysis File zone, click on Select
.
In the dialogue window, choose the corresponding F0 analysis file.
If the F0 was performed on the current open sound, and open, this file will be selected by default, even if not saved.
Multiplication factor of the F0 to determine the harmonics detection bandwitdth.
With a 0 factor, the analysis cannot yield any result.
With a 1 factor, the bandwidth is equal to the F0.
Use the slider or enter a value.