transformers
Word-level timestamps broken for short-form audio
#30325
Merged
Overview
Commits
10
Changes
View On GitHub
Commits
  • force chunk_length_s in AutomaticSpeechRecognitionPipeline
    kamilakesbi committed 2 years ago
  • compute num_frames even when stride is None
    kamilakesbi committed 2 years ago
  • add slow tests
    kamilakesbi committed 2 years ago
  • fix test
    kamilakesbi committed 2 years ago
  • Update src/transformers/pipelines/automatic_speech_recognition.py
    kamilakesbi committed 2 years ago
  • Update tests/pipelines/test_pipelines_automatic_speech_recognition.py
    kamilakesbi committed 2 years ago
  • add input validation
    kamilakesbi committed 2 years ago
  • Merge branch 'whisper_large_fix_word_level_timestamps' of github.com:kamilakesbi/transformers into whisper_large_fix_word_level_timestamps
    kamilakesbi committed 2 years ago
  • fixup
    kamilakesbi committed 2 years ago
  • small fix
    kamilakesbi committed 2 years ago
FAQ Terms Privacy Refunds Impressum
Loading