Quickstart

After installing fasttrackpy, you can use the fasttrack commandline executable. For a full list of all settable options, see All FastTrack Options

Audio File Only Processing

You can process either a single audio file or a directory of audio files with fasttrack audio. To use the fasttrack defaults, just provide fasttrack audio with the path to the input file or directory, and an output file or directory.

fasttrack audio --file assets/audio/ay.wav --dest assets/audio

This will save the formant tracks for this single wav file to assets/audio/ay.csv/.

To process an entire directory of files, instead pass the directory to --dir.

fasttrack audio --dir assets/audio/ --dest assets/audio

This will save the formant tracks for each audio file with the same file name to the output directory.

Audio File + TextGrid Processing

You can process an audio file + textgrid combination with fasttrack audio-textgrid. The defaults assume you are passing a textgrid with a Word and Phone tier from forced alignment, but this can be overriden.

fasttrack audio-textgrid --audio speaker.wav \
    --textgrid speaker.TextGrid \
    --dest output

Adjusting the assumed TextGrid format

If your textgrid is not formatted as force aligned textgrid, you need to pass --entry-classes the value SequenceInterval and --target-tier the name of the tier you want to analyze.

fasttrack audio-textgrid --audio speaker.wav \
    --textgrid speaker.TextGrid \
    --dest output \
    --entry-classes SequenceInterval \
    --target-tier phones

Adjusting the target intervals

You can configure fasttrack to only process certain intervals by passing --target-labels a regular expression matching values in your textgrid. For example, if you have segmented only the vowels you want to analyze, you can pass --target-labels the value ".".

fasttrack audio-textgrid --audio speaker.wav \
    --textgrid speaker.TextGrid \
    --dest output \
    --entry-classes SequenceInterval \
    --target-tier phones \
    --target-labels "."

Processing a corpus

If you have a corpus of audio + textgrid pairs, you can process them with fasttrack corpus. You can either write the output to one large file, or have it separated by original file name & speaker with the --separate-output flag.

fasttrack corpus --corpus data \
    --dest output \
    --separate-output

Using a config file

There are many possible options to set for both fasttrack audio and fasttrack audio-textgrid. Instead of passing them all at the commandline, you can pass either command a yaml config file.

# config.yml
min_max_formant: 3000
max_max_formant: 6000
nstep: 10

fasttrack audio --file ay.wav \
    --dest output \
    --config config.yml