vm6eji6m4/whisper-chinese-pro

Chinese (Mandarin/Japanese/Korean) + English podcast transcription with speaker diarization, word-level timestamps, hotword injection. Whisper large-v3-turbo + pyannote 3.3.

Public
4 runs
GitHub
Playground API Examples README Versions
  1. Author
    @vm6eji6m4
    Version
    cuda12.1-python3.11-torch2.3.1-X64
    Commit
    ad583bacb94e2df503caae09a263652f3662ad75

    075c4b06

    Latest
  2. Author
    @vm6eji6m4
    Version
    cuda12.1-python3.11-torch2.3.1-X64
    Commit
    a1c2fc6801902cbeba4df0fd61b3e1bf8aa6cb6a

    fd746ece

  3. Author
    @vm6eji6m4
    Version
    cuda12.1-python3.11-torch2.3.1-X64
    Commit
    267c41ef30047cac4c625ff5017d92048ce7631f

    002c2c3f

  4. Author
    @vm6eji6m4
    Version
    cuda12.1-python3.11-torch2.3.1-X64
    Commit
    aaf15cb9ebc7877f7581613a902a9dcc0e8eb150

    f52562c1

  5. Author
    @vm6eji6m4
    Version
    cuda12.1-python3.11-torch2.3.1-X64
    Commit
    0072ccf99effce86fe334bdbfc256244e054591d

    d488c22c