Commits · master · SexHackMe / airtanscropt · GitLab

11 Dec, 2025 24 commits
- Add offload_buffers=True for Qwen-Omni model loading · fbe5c3fa
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  fbe5c3fa
- Fix system content format for Qwen-Omni processor · e2ef1fe8
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  e2ef1fe8
- Add system prompt to Qwen-Omni conversation · 26701b10
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  26701b10
- Add --whisper option, use Qwen-Omni by default · 5d5500e4
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  5d5500e4
- Revert to Whisper for transcription to fix execution errors · 1cfee90a
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  1cfee90a
- Extract only the assistant's response from the generated text · 7d3a3238
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  7d3a3238
- Set images=None, videos=None to avoid image processor error · 407226d7
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  407226d7
- Remove qwen_omni_utils import and set audio inputs directly · 252525b0
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  252525b0
- Move qwen_omni_utils import after loading processor to enable remote code · 6b2dfd39
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  6b2dfd39
- Add trust_remote_code=True to load qwen_omni_utils · 5a419365
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  5a419365
- Update to use Qwen2_5OmniForConditionalGeneration and correct inference format · 54eb31b3
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  54eb31b3
- Add trust_remote_code=True for Qwen2.5-Omni-7B loading · 417e1b19
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  417e1b19
- Implement Qwen2.5-Omni-7B using AutoModel as per Hugging Face documentation · 994dbc48
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  994dbc48
- Final version using Whisper for transcription, Resemblyzer for diarization · 85466098
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  85466098
- Fix model loading to use AutoModelForCausalLM for Qwen2.5-Omni-7B · 1439038f
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  1439038f
- Add torchvision dependency for Qwen2.5-Omni-7B · 0d6f3fcb
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  0d6f3fcb
- Add pillow dependency for Qwen2.5-Omni-7B · 82c1d69b
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  82c1d69b
- Use Qwen2.5-Omni-7B for transcription with 4-bit quantization · cbedaa32
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  cbedaa32
- Switch to Whisper for transcription, remove Qwen-Omni due to repo not found · 3c9ba76d
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  3c9ba76d
- Add setup · 7f6696b6
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  7f6696b6
- Replace pyannote with resemblyzer for token-free diarization · ef3ace79
  
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  ef3ace79
- Make HF_TOKEN optional for pyannote models · 460eaa23
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  460eaa23
- Add README with setup and usage instructions · 45156b52
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  45156b52
- Initial commit: Audio transcription app with Qwen-Omni-7B and speaker diarization · 633bba14
  Stefy Lanza (nextime / spora ) authored Dec 11, 2025
  
  633bba14