Scripts and other research products

NOTE: I've recently moved to using Open Science Framework and github to share intermediate research products. Please check out my profiles there, or links next to each publication in my pub-list for code/data relevant to each publication.

Warning: These scripts are not for download & use!
! They are here for those looking for ideas on how to write their own. If you have criticisms, comments, ideas, etc, please let me know!

Praat
 
Lots of scripts are listed in Joe Toscano's archive.

Acoustic analyses

Measure formants more reliably across talkers
    -- This is an implementation of Escudero et al (2009, JASA) proposal for automatic ceiling optimization. It gets F1 and F2 from a range of ceilings.  for wavs whose vowels have been coded in textgrids.
    -- You would then run this accompanying R script to select from all ceilings that for which the within-category variance in F1 and F2 are minimized. 

V-lover: lists sound files (which can be long) in a subdirectory, and searches for corresponding textgrids that contain tagged vowels, and outputs correlates of vowel quality and nasality (pitch, duration, F1-2 -- you should add F3 for some vowels!!--, F1 bandwidth, first and second nasal poles amplitude). Used for the analysis of vowel nasality and tenseness, tested on a corpus of infant-directed and adult-directed American English and Quebecois French recorded in the lab, and the Hillenbrand corpus. 
    -- An old version which did not use ceiling optimization -- simpler, but fares much, much worse. 
Acknowledgments inside the scripts. 

All-lover: requires textgrids and sound files (which should be short), as well as other ancillary files, and, maximally, outputs vowel (F1-3), consonant (VOT for stops; F2 and frication moments for fricatives), and phrase information (speech rate, duration, pitch characteristics), depending on the input.

Varia
- Concatenation codes: 1, 2, 3, ... examples of scripts I use to generate sound files or randomizations for experiments
- Shift all end boundaries to the right
Draw spectrogram and pitch of sets of files onto eps files. I had transitive and intransitive sentences, so the name of the files -- e.g. intrBBaf_1, intrBBaf_1, transBB_1, transCH_1-- indicates which are the sets. For a list of files whose names are not identical in their first 8 characters, it'll store one eps file per sound file.

Phonetic coding instructions
These have been developed as aid for mini-training sessions at Purdue, and could therefore be more underspecified than total newbies would like. If so, I'm happy to hear where I stopped short!



R
Plot vowel space triangle. Another version. Scatterplot vowels per vowel identity per talker group.

Generate summary tables from the output files of the v- and all-lover scripts above. 

Links to tutorials etc on linear mixed models.


Psyscope

Adult phonotactic training: Play a sound, get a response; draws from training list only in blocks 1-3; from both training and test lists in blocks 4-6 (with twice as many training as test); and only test in blocks (different question as in testing). The design comes from a study by Amelie Bernard, Kris Onishi, & Amanda Seidl that will be surely published soon.

Fixed movie presentation, with serial marks being sent out, and infant-controlled paired comparisons interspersed. This one was built from one used by Judit Gervain by Natalia Egorova and me, with help from Luca Filippin.


Latex/BU sample
After Amanda Seidl and I spent several whole days getting the new BUCLD proceedings formatting right, I put together this so that others do not have to waste precious research time on prettying up the paper to their standards. Note from 2012: It seems the format has changed again. This one may work better.

Comments