[praat-users] Word detection

[[e-mail] [praat-users]]

Dear Praat users,

I am a research student at the University of Toronto working on a psychiatric study for which I need to analyze thousands of audio files on Praat.

I was wondering if there is a way to highlight a specific section of the audio (a sentence somewhere in the middle of the file) and analyze the speech parameters for that section only.

I don’t want to crop the files one by one to get that specific sentence, which is in all the files. Is there a way that Praat can recognize the wavelength pattern of that section and run the analysis on it?

I would appreciate it if you could help me with this.


Golsa

[José Joaquín Atria [e-mail] [praat-users]]

Post by ***@gmail.com [praat-users]
I was wondering if there is a way to highlight a specific
section of the audio (a sentence somewhere in the middle
of the file) and analyze the speech parameters for that section
only

it depends on what you mean by "analyze the speech parameters". You
can get the "voice report" for that section by clicking on `Pulses` >
`Voice report`, which will only take that selected part into account.
If you the analysis you are running is in some other command, or in an
external script, then it will depend on the command or script. And
although you are not keen on it, the last possibility is to extract
that part of the audio and run the analysis on that fragment, as a
separate Sound.

So, yes: there are ways to analyse only specific sections of a recording.

Post by ***@gmail.com [praat-users]
Is there a way that Praat can recognize the wavelength
pattern of that section and run the analysis on it?

Not sure what you mean here. You mean, select a section of a file, and
get Praat to automatically select other sections, in the same or in
other files, which somehow "match" the sound you selected (which is
what your subject line implies)?

Then no, there's no way to do this out of the box.

The normal way to do this is to label the sounds yourself (either
using some sort of ASR or annotating it yourself) and then use those
labels to match what you need to process.

José Joaquín Atria

www.pinguinorodriguez.cl

Post by ***@gmail.com [praat-users]
Dear Praat users,
I am a research student at the University of Toronto working on a
psychiatric study for which I need to analyze thousands of audio files on
Praat.
I was wondering if there is a way to highlight a specific section of the
audio (a sentence somewhere in the middle of the file) and analyze the
speech parameters for that section only.
I don’t want to crop the files one by one to get that specific sentence,
which is in all the files. Is there a way that Praat can recognize the
wavelength pattern of that section and run the analysis on it?
I would appreciate it if you could help me with this.
Golsa