I am new to Praat and tryingt to get the selected region in Praat script using the the amplitude values i.e. when amplitude gets larger than a threshold then I should start capturing the region and stop when it falls down. enter image description here

What I know is that:

There is something called Shimmer (amplitude perturbation) which measures the difference in amplitude from cycle to cycle. Once again, this is a useful measure in speech pathology, as pathological voices will often have a higher shimmer than healthy voices (although again, both healthy and unhealthy voices will have some shimmer).

I have also read this link but couldn't get anything to measure the shimmer in praat script.
Someone please guide me in this regard.

  • Did you also look at other parts of the Praat manual? This seems to directy address shimmer in scripts: fon.hum.uva.nl/praat/manual/… – Jeremy Needle Oct 4 '16 at 13:01
  • Yeah I looked at this one, I want to write all the shimmer values across time units in a text file but I couldn't do this because of lack of knowledge. – Itban Saeed Oct 5 '16 at 12:34
  • That's not hard to do. If you still want to try that you can start a new question. As for this question, does my answer not work for you? – jja Oct 7 '16 at 11:35
  • well, that seems helpful but unfortunately I couldn't find the proper way to get the To TextGrid data and in the mean time we are working on another approach to do the job. – Itban Saeed Oct 9 '16 at 12:17
  • Are you looking for something that would e.g. identify "a person is talking" vs. "nobody is talking"? What is the threshold you're referring to, and what is the duration of the difference in question? That is, why do you care (relevant to getting what you are really looking for). – user6726 Oct 10 '16 at 5:25

The variations shimmer refers to are much smaller than what it looks like you are after.

What you seem to want is most easily achieved with the To TextGrid (silences)... command for Sound objects. You'll have to adjust the intensity thresholds, but it should do what you want.

| improve this answer | |
  • thank you so much for the answer :) I am now working on TextGrid but I found that number of sounding regions and the accurate start point depends on the silence threshold value (i.e we have to change that for different environments) Can you please recommend a generic way to calculate the silence threshold ? – Itban Saeed Nov 3 '16 at 5:53

I suggest reading the Voice tutorial in Praat, which you will see a link to if you check help searching for "pulses". There is a display, 'pulses', which marks "voiced pulses". There are two kinds of things with "big amplitude", voiced sounds, and noisy (voiceless) release bursts. If you were dealing with clicks, those bursts would be really loud, but they are fairly short, so you basically want quasi-period high amplitude sections, plus (good luck with that...) not 60 cycle hum. You should expect there to be a few problems if you're dealing with English and you want to identify [g], because [g] especially can be phonetically voiceless. Apart from that, a pulse listing sounds like what you want.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.