Faster Whisper implementation using optimized CTranslate2 models.
GPU execution requires cuBLAS and cuDNN libs.
Last included commit: #149
Includes PR #163
Some new stuff:
no_repeat_ngram_size arg.
Fix memory leak when batch processing.
Purfview/whisper-standalone-win
faster-whisper
Whisper-Faster r149.1
on GitHub
latest releases:
Faster-Whisper-XXL, Whisper-Faster-XXL
13 months ago