If you would like to run the open-speech-application locally please follow the instructions below. In order for this to work you need to have Python installed on your local machine as well as the python package manager pip.
We have forked a version of Pete's Open-Speech-Recording application to use only local memory which you can use to record your own datafiles. If you'd like a more robust version that you want to share with others to aid in your data collection (and therefore use a google storage buckets which may cost some money) please see Pete's original project. You can setup and use our version of the app as follows.
- Clone the repository and install the only requirement, flask:
git clone https://github.com/tinyMLx/open-speech-recording.git
pip install flask
- You can then run the server locally (from within the open-speech-recording folder) by running:
cd open-speech-recording
export FLASK_APP=main.py
python -m flask run
-
Then open the link provided in the terminal in a web browser to run the application. Make sure to run the application in a private or incognito window which avoids any cacheing issues. Also we've found that the app works best when using Chrome so if you are having issues with another browser please try it in Chrome.
-
The application can then be run as specified in the edX course instructions. Once you are done recording, click to download the audio files. You will find that instead of downloading the files individually or via a zip, you now have a large collection of
.oggfiles in your open-speech-recording directory. -
You can now safely close the flask app from the command line by running
cntrl+c(orcmd+c) and can now proceed to formatting your audio files and then training your model in Colab.