Audacity Forum

More improvements. Used more Nyquist functions to avoid doing much math in Lisp. I can now segment some selections of my audiobooks in about 0.5 seconds calculation per second of audio, about a tenfold improvement over the previous. Improving the quality of the results I get is another matter for ex...

... and I couldn't sleep until I tried that out! New code, new picture with a new second track which is the result of the new defaults on the same sound. New processing is done in Nyquist-y ways avoiding new inner loops in Lisp. Need to consider how to eliminate my old inner loops in lisp. It's afte...

The same but with fft window size of 1024 instead of 256. Much smoothing out of noise in the graphs. Notice not so much the levels as the steep areas. Now I'm thinking a program that finds those steepnesses and labels the areas between them could do a pretty good job of segmentation of speech. I als...

Here's a cute hack... generate a graph as a waveform in an effect. Duplicate some audio and apply an effect to it. Here is me saying "a long time ago in a galaxy far, far away" in a monotone. And graphs showing the frequencies that mark the 75th, 50th, and 25th percentiles of the power spectrum, wit...

Here's a more elaborate version of the tool, that lets you fix the "percentile" parameter and see varying results for different frequencies. And here is me babbling "Tha-fa-sha-sa-kha-ha" in spectrogram, and the results of this version of the tool, default settings. It does a pretty good job of slic...

What did you say? I'm not that good at reading waveforms! As I said, the right threshold for separating the voiced stops should indeed be context dependent but I hope sibilants might be easier. Here's a few results I've seen with a few little examples with lengthened sibilant sounds, perhaps these m...

Hm, I have a little test sample of my own voice in a certain register, containing only vowels and voiced stops, for which the indicated numbers worked to isolate voiced stops. A proof of concept. But the correct frequency wouldn't be fixed... I picked out some other less contrived recorded text and ...

Play with it, kozikowski, and see first if my simplistic sorts of criteria can be good enough to demarcate sibilants. My toy is still very green but as described I think I can isolate m, n, ng pretty well, at least for my own adult male voice: the right frequency to use may depend on the speaker of ...

Hm, I think the placement of the band is varying for me with the zoom level. In a bad, misleading way, actually, sometimes with 173 not even inside the band! But it seems to prefer to put 173 at the bottom of the band if I zoom the scale to put 1722 at the top.

What did you say and what were your settings? There are several dials to fiddle with as you can see. I see you got a lot of extraneous labels in your pauses, and lots of little labels. Things to work on. The criteria it applies go by where the median or other percentile of the power spectrum falls, ...