Chat GPT can write #Praat scripts. Damn.

I’ve published a new python package to make navigating and working with force aligned TextGrids easier! #python #praat #phonetics

PyPi: https://pypi.org/project/aligned-textgrid/
Docs: https://alignedtextgrid.readthedocs.io/en/latest/

The process works! Now to do the whole batch and see what the group result is.

Back in the 2000's, as an undergrad I did manual identification and labeling of acoustic markers of speech in #Praat. 10 to 20 hours a week for 2 years measuring German and English consonants. It is incredible what is possible now using computers.

VoiceLab v1.3.1 has been released.

Check out the changes and get your copy here: https://github.com/Voice-Lab/VoiceLab

VoiceLab is free, open source software that provides reproducible, automated voice analysis, visualizations, and manipulations.

#Praat #Librosa #VoiceSauce #AcousticAnalysis

@psychology
@psycholinguistics
@linguistics
@cogsci
@cognition

I'm preparing another release of VoiceLab. Just need to compile on Windows.

Anyways the new release modifies Cepstral Peak Prominence to measure only voiced frames (with an option to turn this off for whatever reason), fixes for energy where it wouldn't work on compiled versions, and API updates for command line interface users.

#Praat #AcousticAnalysis #AcousticPhonetics

Wait wait. Turns out I get the "bad" file-save-menu when viewing a sound -> File -> Save selected sound as WAV, but when opening or saving from the Objects window, I get the "good" file-save-menu. #Praat just has both! ARGH

reunited with this mug on a day when I was suitably spending a lot of time on #Praat

#phonetics #PhoneticLife

reunited with this mug on a day when I was suitably spending a lot of time on #Praat

#PhoneticLife #phonetics

#logo #fail The "Managed Software Center" app provided by my employer showed a weird icon in its "Educational" category of apps it can install. It was so weird, I needed more info. I learned it's for #Praat, a speech analysis program that's been around for 30+ years. The logo has been through an evolution, but it still looks really WEIRD.

The #Reddit thread about the logo is pretty funny: https://www.reddit.com/r/logodesign/comments/rmzh66/the_old_comically_atrocious_logo_for_praat_a/

I think someone needs to dip into some clipart.

@nickfisherau Nice. Did you publish any details about your alignment code?
For post-edit - maybe #Praat is easier for #speechtech?
I recently did phone alignment tool as a gift to Czech #phonetics and minimizing dependencies (and install hassle), I finally went without anything like Kaldi, using just very basic #Pytorch for NN AM training and then alignment. I guess going zero-up was really easier than decomposing some big thing like Whisper.
https://github.com/vaclavhanzl/prak

@stefanowitsch @linguistics #Python for dealing with #corpora, #R for #stats, #LaTeX with #knitr for transparency. I use #PowerShell a good amount too when I need to batch process files (e.g., convert a bunch of stereo audio files to mono). Sometimes I like using PowerShell over Python for text normalization over many files just because I've always found the file management syntax in Python clunky (e.g., removing diacritics). Also, let's not forget about #Praat scripting for #phonetics stuff

Useful: A Practical Guide to Calculating Cepstral Peak Prominence in Praat by Elizabeth S.Heller Murray; Andie Chao & Lauren Colletti

"CPP is the recommended acoustic measure of dysphonia, replacing previously relied upon perturbation measures (jitter and shimmer)."

https://osf.io/yvp4s/

Praat CPP plugin here - Cepstral Peak Prominence - Praat with and without voice detection

https://osf.io/t5hrv/

with some videos on how to add the plugin to Praat.

#Cepstral #Peak #Prominence #Praat

I've begun a complete rewrite of the VoiceLab backend. It's enough hacking. Time to start from scratch. I won't change the measurements or anything... Just the GUI and the data pipeline to accommodate multiprocessing, and in doing so, I'll tidy up the codebase.

While I'm developing (will likely take many months), have a look at my latest release:
https://github.com/Voice-Lab/VoiceLab

#Praat #AcousticAnalysis #VoiceSauce #Python #PySimpleGUI #VoiceAnalysis

@aart_vandenberg ik vind het sowieso al een prestatie om naar die #praat programma's te kijken. Ik kan het in ieder geval niet 🤔

@stefanowitsch @linguistics I once just *had to* use #Praat to draw some pictures, but as Praat #scripting is horrible I ended up writing a #Python script to generate a Praat script that finally created those pictures. #justanidea

Back to transcribing speech in #Praat after ~5 years!

in that vein, there are a lot of pretty extensive #tutorials on #Praat #scripting, including these two websites:

* https://praatscripting.lingphon.net
* http://praatscriptingtutorial.com/

there are also a number of slide-decks out there if you google "Praat scripting tutorial" — those might be better for some situations than these more comprehensive tutorials.

if you can't find the answer there, there's a very active Praat message board:
https://groups.io/g/Praat-Users-List

let's start with the #scripting resources that can be found on Praat's own website

* a scripting tutorial: https://www.fon.hum.uva.nl/praat/manual/Scripting.html
* a list of functions: https://www.fon.hum.uva.nl/praat/manual/Functions.html

you'll notice that list doesn't have any of the menu-item commands… for how to call those functions, check out this page http://praatscriptingtutorial.com/findingCommands

(this relates to a general point that gets brought up by #Praat scripters: Praat's own documentation could be much better)

the other day on the twit site, some questions were raised about how to do #Praat #scripting using Points tiers, and some things to share from that thread

there are some very good resources for doing #Praat #scripting, and also some ways to interface with Praat using scripting languages like Python and R