Skip to content

Conversation

@careo-h
Copy link

@careo-h careo-h commented Nov 13, 2025

  • Make device and compute_type parameters configurable in FasterWhisperASR
  • Add model_size_or_path parameter support for custom models
  • Fix hardcoded CUDA device causing RuntimeError on AMD systems
  • Remove pyannote_vad import to avoid NumPy 2.0 compatibility issues
  • Maintain backward compatibility: defaults to cuda+float16 for NVIDIA users

Fixes runtime errors on AMD GPUs and CPU-only systems:
RuntimeError: CUDA failed with error CUDA driver version is insufficient

Tested on: AMD Ryzen 12-thread, Ubuntu 24.04, Python 3.12

Usage for CPU-only:
--asr-args '{"model_size_or_path": "turbo", "device": "cpu", "compute_type": "int8"}'

- Make device and compute_type parameters configurable in FasterWhisperASR
- Add model_size_or_path parameter support for custom models
- Fix hardcoded CUDA device causing RuntimeError on AMD systems
- Remove pyannote_vad import to avoid NumPy 2.0 compatibility issues
- Maintain backward compatibility: defaults to cuda+float16 for NVIDIA users

Fixes runtime errors on AMD GPUs and CPU-only systems:
  RuntimeError: CUDA failed with error CUDA driver version is insufficient

Tested on: AMD Ryzen 12-thread, Ubuntu 24.04, Python 3.12

Usage for CPU-only:
  --asr-args '{"model_size_or_path": "turbo", "device": "cpu", "compute_type": "int8"}'
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant