![Gitter](https://badges.gitter.im/Join Chat.svg)
A listing of publicly available audio samples of a large number of different languages.
The Wide Language Index is an attempt to curate a set of examples of a wide variety of languages from public podcasts. The index itself contains a listing of samples and information about them. The audio files themselves are not bundled in, but can be downloaded as needed.
The full dataset is meant to be downloaded from OS X or Linux. If your system meets the required dependencies, fetching examples of different languages should be as simple as running: make env && make fetch
.
To use the tools that accompany this dataset, your computer should have installed:
python3.4
pip
virtualenv
make
Run make env && make audit
to check if the tools are working.
The index/
folder contains references to publicly accessible audio samples for which the principal language has been pre-determined. Each sample is represented by a JSON file containing, at minimum:
language
: the ISO 693-3 code for the languagemedia_urls
: the URL of the raw sample filesource_name
: what venue or outlet published the audio filetitle
: the title of the podcast or sampledate
: the date of publication of the samplechecksum
: an md5 checksum of the language
It may also contain the optional fields:
source_url
: the URL of the page containing the podcast, for contextdescription
: a description of the contents, in English or in the source language
Each sample's file is named as <language>/<language>-<checksum>.json
.
A helper script exists for fixing samples, but it will only work with access to the underlying S3 bucket.
python src/recode_sample.py index/<language>/<language>-<checksum>.json <new-language>
This index is provided under the Creative Commons Attribution-NonCommercial 4.0 International License. In short, you are free to use this data non-commercially as long as you provide a reference back to this project. If you have a use case where this would be onerous, please send me an email.
See Contributing.
Lars Yencken lars@yencken.org