Using Praat for audio annotation and acoustic analysis

The Goals of this Tutorial:

After this workshop, you should:

1) Be comfortable playing audio sounds with Praat and creating TextGrid files for sound segmentation and annotation.

2) Be able to use visual cues from waveforms and spectrograms to identify segment boundaries.

3) Be familiarized with common acoustic parameters used in the analysis of different kinds of segments.

4) Gain experience running Praat scripts and have the resources to start your own script-writing project.

Before Getting Started...

Download Praat here : https://www.fon.hum.uva.nl/praat/

Download files for this workshop and unzip your download:

Some basic operations

Import an audio file

Praat Objects: Open -> Read from file -> select

View an audio file

Click on the sound object -> View & Edit

Try the following shortcuts:

  • Command + i

  • Command + o

  • Command + a

  • Select a period of time -> Command + x -> Command + v

Play an audio file

Play the visible part vs. play the whole audio

Record an audio file

Praat Objects: New -> Record -> Sound Recorder: Record -> Stop -> Name -> Save to list

Save a recording

Praat Objects: Save -> Save as WAV file

Create a TextGrid

  1. Select the sound object that you want to annotate in Praat, then click on Annotate --> To TextGrid...

This will generate a blank TextGrid file that matches the audio you selected in duration.

2. Specify the tier name(s)

  • The number of words you type in equals to the number of tiers that will be created, with the name of each tier separated by a space.

  • In the dialogue window, the default tier names are "Mary John bell", with "bell" being the point tier. So it means you will create 2 interval tiers called Mary and John and 1 point tier called bell.

  • The difference between a point tier and an interval tier is that an interval tier will give you two timing values for each annotation -- the starting and ending point of an interval that you annotated, while the point tier only gives you the time at one specific point.

  • For the purpose of annotating my data, I am going to create two tiers, one for word and one for vowel.

3. Highlight both the sound and the TextGrid -> View & Edit

4. Create boundaries and annotate sounds

  • Enter: create a boundary

  • option+delete: delete a boundary

5. Save the TextGrid by going to File -> Save TextGrid as text file. Remember to always use .TextGrid as the extension!!