DS-Subtitler

Generate SRT subtitles from libsndfile supported audio files using Mozilla DeepSpeech.

Requires

• deepspeech
• sndfile
• libresample
• rnnoise

Usage

-b beamwidth    Beam width
-m model	Model to use, default is deepspeech-models.pbmm
-s scorer	Scorer to use, default is deepspeech-models.scorer
-w words.txt	Read hotwords from file
-f 2		Frames to use for silence detection for splitting
-l 4		Minimum length in seconds to decode
-r              Use raw audio for DS

Issues

• The code is uggly, very ugly
• 2-channel files are not handled properly yet