This is a Python package that provides various utilities for converting speech to text, force aligning the transcription with the audio, and extracting timing information for a linguistics production experiment that I'm conducting.
The various utilities in this Python package are designed to work with
some data wrangling and data analysis scripts that are not part of this
repository. As such, the way that these utilities interact with the two
different .csv files in the example data folder might seem somewhat
idiosyncractic. Nonetheless, feel free to adapt this Python package for
your own uses if you find it useful. See below for example usage as well
as instructions for using Google Cloud's Speech-to-Text API and
instructions for installing the forced aligner.
I should also note that, while the aligner.aligner.Aligner class
provides functionality for writing Praat .TextGrid files, I do
not make much use of these files myself, so there may be some
undiscovered bugs in the code that generates those files.
And one final note: this repository uses Git LFS to track binary files. In order to clone the repository, you will need to install it.
After cloning the repository, you can install the package locally in a Python virtual environment, like so:
pip install -e .The package provides the following utilities:
to_mp3transcribeextractalign
To see help information, you can run to_mp3 --help after installing
the package (and assuming the virtual environment into which you
installed the package is active).
Example usage:
to_mp3 -l info -f data/ibex_results/example_data_tidy.csv -m data/mp3_files -z data/zip_archivesTo see help information, you can run transcribe --help after
installing the package (and assuming the virtual environment into which
you installed the package is active).
To use this utility, you will need a Google Cloud account. You can sign
up for a free trial. After creating a Google Cloud account, navigate to
https://console.cloud.google.com/speech. If you
haven't already created a project, you will need to do so.
After creating the project, click the "Enable API" button:
Next, navigate to
https://console.cloud.google.com/iam-admin/serviceaccounts,
and select your newly created project. For this project, click "Create
Service Account" to add a service account. Name it whatever you'd like,
and then click "Create and Continue":
Finally, give the Service Account a role so that it can use the Speech-to-Text service. I gave my service account the Editor role from the Basic list:
Then hit the 'Done' button. Next, click on the newly created Service Account, and navigate to the 'KEYS' tab:
Click the 'ADD KEY' dropdown and then click 'Create new key'. Add a .json key, which will be downloaded to your computer.
Now, you can use this utility like so:
transcribe -l info -f data/ibex_results/example_data_tidy.csv -t Transcription -d data/ -m mp3_files -c /path/to/json/credentials/file.json -n 8To see help information, you can run extract --help after installing
the package (and assuming the virtual environment into which you
installed the package is active).
Example usage:
extract -l info -d data -f data/ibex_results/example_data_tidy_transcribed.csv -t Transcription -nTo see help information, you can run align --help after installing the
package (and assuming the virtual environment into which you installed
the package is active).
In order for this utility to work, you must install the Gentle forced aligner and have it up and running on its default port, port 8765.
Example usage:
align -l info -f data/ibex_results/example_data_tidy_transcribed.csv -d data/ -m mp3_files -t transcriptions -g gentle_align -o -p