Logo Lanfrica

Niger-Volta-LTI/audio-corpora-builder

Domaine:

natural language processing

Type de record:

software
Créateur:
Nig
Hôte:
Build large audio corpora in various languages → {Yorùbá, Urhobo, Ẹ̀dó, Èʋe, Ị̀gbò} # audio corpora builder Build large audio corpora in various languages → {Yorùbá, Urhobo, Edo, Èʋe, Igbo} ## Audio Corpora Curate specific language corpora from the wealth of {speech, dialogues, interviews, films, podcasts, etc}, available in good quality on YouTube → ### Quick start example (Yorùbá): * Locate a list of existing playlists, e.g. OrisunTV Iroyin * Alternatively, create a new playlist with a custom set of YouTube videos * Update `yoruba_sources.yml` with the reference to the playlist ```bash $ python3 download_youtube.py --output ./audio --sources yoruba_sources.yml --verbose ``` ## Install dependencies * Python 3.10 or newer * `$ python3.10 -m venv venv` * `$ pip install -r requirements.txt` ### With custom/different playlist files: ```bash python3 download_youtube.py --output ./audio/ --sources urhobo_sources.yml ``` ### With MP3 format: ```bash python3 download_youtube.py --output ./audio/ --format mp3 ``` ## Sanity Test Results ```bash $ python3 -m pytest tests/test_download.py -v ============================= test session starts ============================== ... collected 11 items tests/test_download.py::TestValidateUrl::test_validate_url_valid_playlist PASSED tests/test_download.py::TestValidateUrl::test_validate_url_valid_user PASSED tests/test_download.py::TestValidateUrl::test_validate_url_empty PASSED tests/test_download.py::TestValidateUrl::test_validate_url_none PASSED tests/test_download.py::TestValidateUrl::test_validate_url_missing_https PASSED tests/test_download.py::TestValidateUrl::test_validate_url_with_whitespace PASSED tests/test_download.py::TestBuildYtDlpCommand::test_build_playlist_command PASSED tests/test_download.py::TestBuildYtDlpCommand::test_build_users_command PASSED tests/test_download.py::TestBuildYtDlpCommand::test_build_command_output_path PASSED tests/test_download.py::TestBuildYtDlpCommand::test_build_command_audio_format PASSED tests/test_download.py::TestBuildYtDlpCommand::test_build_comman …