On this page you can find a number of Praat plug-ins and scripts that we frequently use in our lab here in Zurich. These are are to be used with the Praat digital speech analysis software (www.praat.org).  

It would be nice to know who is using these programs and for what. So I very much appreciate a short email with your comments and, if you have, suggestions: volker.dellwo@uzh.ch. If you use the scripts for your publications I would also appreciate if you could send me a pdf of the article for my records. 


Praat Plugins:

Praat Plugins are extensions to the Praat software. You can see them as little programs in themselves. To install a Praat plugin do the following:
durationAnalyzer
Duration Analyzer (version 0.03)
This program analyzes durational characteristics of segments, syllables, consonantal  or vocalic intervals, and many others. All you need is a labelled TextGrid of your speech. Tell the script which tier contains which information, define your free variables (as encoded in the TextGrid name) and off you go.
sentencePresenter
Sentence Presenter (0.05)
This program allows you to present sentence by sentence on a screen with a prompt for speakers to start reading. In addition it creates a TextGrid so that all your sentence recordings are already labelled and can be extracted automatically. You just need a text file with the list of sentences you wish to record.
  • Start your recording (either in Praat or any other device).
  • Start sentence presenter and make sure your have the tone it plays on your recording (it is a calibration tone and you will need it later on). 
  • Record your speaker. 
  • After the last sentence, sentence presenter will stop automatically. 
  • Stop your recording and open it in Praat. 
  • Cut everything from time 0 to the end of the calibration tone in your recording.
  • Open the TextGrid that SentencePresenter saved. Your recorded sentences will all be labelled accordingly. 
Control options:
  • SentencePresenter can either be controlled on the same screen or, if you have an external screen connected to your computer, by a control panel you keep on a different screen.
  • SentencePresenter can either play a countdown or it can play a silent countdown in that it first presents a red screen and then turns into a green screen. The subject should talk after the screen turned green. 
Limitations:
  • Sentence presenter only works with longer sentences right now. Recording short words or other short sounds can lead to synchronization problems. I am working on fixing that.
sasasa pic
Sasasa Delexicalizer (0.05)
Sasasa speech is often used in experiments on speech prosody to expose rhythmic characteristics of a language to the listener without revealing the language identity. In sasasa speech every consonantal interval is turned into an /s/ and every vocalic interval into an /a:/ sound. While sasasa speech is often based on MBROLA synthesis which uses pre-defined voices, this delexicalizer allowes you to your use your own voice.


CV Tier Creator (0.01)
This plugin turns a segment tier into a CV interval tier. It gives you three tiers, (a) a tier containing consonantal and vocalic segments, (b) a tier containing consonantal and vocalic intervals with each interval containing the number of underlying consonantal and vocalic segments respectively and (c) a tier containing consonantal and vocalic intervals.
BFC_experimenter

Binary forced-choice Experimenter
This is a simple interface for obtaining perception data on binary forced-choice tasks (including three levels of certainty). Just upload your stimuli in the 'stimuli' directory and off you go (good for quick demos in the classroom).

 


Praat functions:

Praat functions are simply scripts that carry out a certain task to an object in your list of objects. They typically create a new object based on the object you are executing the script on. You can simply copy and paste the text of these scripts into a script window and run it.

(a) Functions for signal manipulation:

This script takes each peak point between two zero-crossings of a waveform (either positive or negative) and sets it to a chosen constant amplitude value while scaling all other waveform samples between the same zero-crossings proportionally. The result is a signal with a flat amplitude envelope in which the temporal fine structure is preserved. 
This script requires a sound and a TextGrid with an interval tier (e.g. syllabic intervals). It creates a new sound in which all intervals from the interval tier will have the exact same duration. The total duration of the new sound will be the same as of the original sound.
This script turns speech into vocoded speech. You can choose the number of channels for your vocoder and you can choose from a variety of source signals.

This script shifts your frequency spectrum up by a chosen frequency.

This script divides your frequency-domain into a number of bands of a chosen bandwidth and rotates these bands.

This script inverts the pitch line around the average of the total pitch.

This script reduces the number of quantisation levels of a speech signal to 2. All amplitude values below 0 will be set to 0 and all amplitude values higher than 0 to 1.

This script adds two or more different sound objects in the list of objects.


(b) Functions for signal analysis:

This script creates a TextGrid with voiced and voiceless intervals in your speech signal.

(c) Editing functions:

This script adapts the length of a TextGrid to that of a wave file. The new TextGrid can be scaled so that the first label is larger than original. This comes in very handy when applying the same labelling tier to different soundfiles manually.


Speech data:

focusCorpus