k y m a • t w e a k y — the kyma collective || view the current website || February 2015 Archive

Search

How Section

All Sections

Symbolic Sound

TWiki Links

Question (or Task)

I am trying to pitch-track voice, and am looking for suggestions for parameter settings in the PitchTracker?. Since I don't know how it works, and none of the parameters are "hot", it makes fine tuning problematic. What is the effect of using more (or less) detectors? "Confidence" is extremely sensitive - it seems it must remain close to 0.40 - but it also seems to be a function of the number of detectors. I also seem to get better results with "Emphasis" set very low - contrary to what the help info says.

Also, what about pre-conditioning the signal with compression, or band-limiting to only the frequencies of interest? Will this improve tracking?

I am giving a lecture/demo on vocoders in a couple of days and need a good tracking Sound. Any suggestions or advice?

-- BillMeadows - 19 Nov 2003

Solution(s)

I just tried a few things and found that high pass filtering the voice helps with the tracking. I used 80 hz as cutoff frequency, and fourth order.

Another thing would be to set the MinFrequency and the MaxFrequency parameters of the FrequencyTracker to the smallest range that would capture the frequencies of the voice.

-- KurtHebel - 19 Nov 2003

In my experience, changing the min and max frequencies has the biggest influence on the tracking. I usually run the input through the FrequencyEstimator? first (see Sound Library>Effects>Frequency & Time Scaling>Frequency scaling).

-- CarlaScaletti - 19 Nov 2003