This is an application for practicing a foreign language using a variant of the "dictogloss technique". There are different reasons/approaches with this technique, but I found it incredibly useful for understanding people speaking in real Swedish. I believe that it is often also used as a means of teaching grammatical constructs as well. The flow of the application is as follows:
- A link to an MP3 is provided, which is downloaded and split into segments of 30s each.
- The segment is played over and over, and the text is written into the text box on the page.
- The answer is submitted, and is then compared with the transcript of the segment, obtained using the Google Speech API.
At present, this app is not hosted anywhere. It is strongly suggested that you only bind to 127.0.0.1, as
otherwise anyone will be able to use your API credentials (and therefore your associated billing account) to transcribe
any speech that they like.
To obtain the transcript of the application and compare with an answer, this application will require a Google Cloud account, with the Speech-to-text enabled.
NOTE: The Google API is free for the first sixty minutes of audio transcribed, but after that will cost as documented on the Google Cloud Speech-to-Text Documentation.
The app can probably run without that, if you only need to download/store a file, and split into segments.
It has been tested on Ubuntu 18.04 with Python 3.6. It will probably work on Windows, and might work with Python 2 (unlikely) but no support will be given for those.
To set up the Google Cloud Speech API, follow the instructions for authentication.
This app is stored in the Docker Hub at https://hub.docker.com/r/huwf/dictogloss and can be retrieved as follows:
docker pull huwf/dictogloss
You will need to mount the API key to the docker container. The code is stored at /usr/src, so the easiest is to mount
the file from the host to /usr/src/API_KEY.json.
Depending on your setup, you will probably want to mount the mp3 directory to host, and the sqlite file (if appropriate).
By default this is database.db.
It runs using the Flask development server on port 5000, so can be run as:
docker run -v $(pwd)/database.db:/usr/src/database.db -v $(pwd)/API_KEY.json:/usr/src/API_KEY.json -v $(pwd)/static/mp3:/usr/src/static/mp3 -p 127.0.0.1:5000:5000 huwf/dictogloss
Alternative settings for the Flask server can also be set as arguments.
Install the dependencies using the command:
pip install -r requirements.txt
After you've saved the JSON file you get from setting up the requirements, set the GOOGLE_APPLICATION_CREDENTIALS environment
variable to the location you have saved the file, e.g.,
export GOOGLE_APPLICATION_CREDENTIALS = API_KEY.json
To run the application:
export FLASK_APP = app
python -u -m flask
Limited customisation can be done through environment variables:
- Flask environment variables can be configured in the usual way according to the Flask documentation
- You can change the database connection string to use a different server by modifying the
DB_CONNECTION_STRINGenvironment variable.