rajivpoddar / text_audio_align

Generate text and audio splits from strings successfully aligned by Gentle

Dump of files related to text and audio alignment, language models, and ASR data generation/testing/prediction.

Aligner and manifest generator

Generate text and audio splits from strings successfully aligned by Gentle, and write a manifest file of the generated files.

aligner

with asr_data_gen.py and aligner.py moved to /PATH/TO/gentle/,

$ cd gentle
$ python asr_data_gen.py FILE_ID

manifest generator

• to be used with pytorch DS2

$ python pytorch_manifest.py (optional: --files_dir /PATH/TO/WAV_AND_TXT/DIRS/ --out_file /PATH/TO/MANIFEST_FILE/)