CommonVoice Bundler

Script for bundling Common Voice (https://voice.mozilla.org) clips by language.

What it does

1. Query database for all clip data
2. Download all those clips from an S3
3. Anonymize clips client_id and filename (called path)
4. Upload a tsv file with all the anonymized clip data
5. Put clips into archives by language and upload it to (a different) S3

How to run it

1. Install node (>= 8.3.0)
2. Install yarn
3. Install CorporaCreator
4. Install mp3-duration-sum
5. git clone git@github.com:Common-Voice/common-voice-bundler.git
6. Override the keys defined in config.js with a config.json in the same dir
7. yarn
8. yarn start