Written and maintained by CASRAI Editorial Board
Last updated
Praat is free, open-source software for analysing, annotating and manipulating speech, developed at the University of Amsterdam. Researchers use it to measure pitch, formants and intensity, to view spectrograms, and to label recordings. This page explains what it does, where each task is documented in its official manual, and what to record so an analysis can be repeated.
What is Praat software used for?
Praat is used for phonetic analysis: measuring and visualising properties of recorded speech. The program’s home page describes it as “doing phonetics by computer” and lists speech analysis (spectrograms, pitch, formants, intensity), speech synthesis, labelling, statistics and scripting among its capabilities. It is used in linguistics, including sociolinguistics work that depends on acoustic measurements, and in other fields that study the voice.
Who develops Praat, and is it free?
Praat is developed by Paul Boersma and David Weenink, with Anastasia Shchupak, of the Phonetic Sciences group at the University of Amsterdam. It is free to use. The manual’s acknowledgments page states that the whole of Praat is distributed under the GNU General Public License, version 3 or later; some bundled third-party components carry their own licences, which that page lists separately.
The official site offers a separate download page for each platform: Macintosh, Windows, Linux, Raspberry Pi and Chromebook are listed on the home page, along with several older Unix systems. Praat is a desktop program; this page makes no claim about web or phone versions.
What does the word “praat” mean? Is it Dutch?
Yes. “Praat” is Dutch for “talk” (Wikipedia gives the Dutch pronunciation as [praːt]). The program was first released in 1992 according to the same article. The official pages read for this guide do not explain why the name was chosen, so no origin story is given here.
Where to start: the manual’s tutorial chapters
The official manual includes an introductory tutorial whose chapters map directly onto common research tasks. This mapping is the quickest route into the documentation.
| Task | Manual tutorial section |
|---|---|
| Get audio into Praat (record, read from file, create from a formula) | Intro 1: How to get a sound |
| Save and view a sound | Intro 2: What to do with a sound |
| Spectrograms and spectral slices | Intro 3: Spectral analysis |
| Fundamental frequency (pitch) tracking | Intro 4: Pitch analysis |
| Vowel and resonance measurements | Intro 5: Formant analysis |
| Loudness-related measurement | Intro 6: Intensity analysis |
| Labelling segments of a recording | Intro 7: Annotation |
| Changing pitch, duration, intensity or formants | Intro 8: Manipulation |
Scripting is documented in a separate tutorial. The Scripting page defines a script as a text made of menu commands and action commands, run “as if you clicked on them”. Its sections include “Your first scripts”, “How to script settings windows”, “Object selection”, “Scripting the editors”, “Controlling Praat from another program” and “Turning a script into a stand-alone program”.
Strengths and limits
- Strengths: it is free under the GPL, runs on the major desktop systems, covers analysis, annotation and manipulation in one program, and can be automated through scripts so that the same measurement is applied to every file.
- Limits: Praat is a desktop program with its own interface and scripting language, so batch work means learning its script syntax; and the measurements it returns depend on the analysis settings you choose, so those settings should be reported with results.
Making a Praat analysis reproducible
Because settings change results, record the Praat version you used , keep your scripts with the data, and report analysis parameters alongside your results. Scripting is the practical way to do this, since a saved script documents every step. For annotation of speech for qualitative or corpus work, see transcription conventions in qualitative research; for a shared repository of annotated language data, see TalkBank and CHILDES; for the underlying concepts of analysing waveforms, see what is signal processing.
FAQ
What is Praat software used for?
Analysing and annotating recorded speech: pitch, formants, intensity, spectrograms, labelling and resynthesis-style manipulation.
Is Praat free to use?
Yes. The manual states Praat is distributed under the GNU General Public License, version 3 or later.
Does “praat” mean something in Dutch?
Yes, it is the Dutch word for “talk”.
Which platforms does Praat run on?
The official home page lists Macintosh, Windows, Linux, Raspberry Pi and Chromebook, with a download page for each.
Can Praat be automated?
Yes. The manual has a dedicated scripting tutorial, including controlling Praat from another program.
Sources
- Praat: doing Phonetics by Computer, home page (read 2026-11-07)
- Praat manual: Intro tutorial (read 2026-11-07)
- Praat manual: Scripting (read 2026-11-07)
- Praat manual: Acknowledgments and licence (read 2026-11-07)
- Wikipedia: Praat (meaning of the name and first release year; read 2026-11-07)
