O
Whisper.cpp
by openaiPublicmit
README.md
Whisper
Whisper is a state-of-the-art automatic speech recognition (ASR) and speech translation model family developed by OpenAI, published under the openai organization on Hugging Face.
Links
- Upstream model card: https://huggingface.co/openai/whisper-large-v3
- Upstream GitHub: https://github.com/openai/whisper
Model Facts
- Task: Speech-to-Text (ASR) & Speech Translation
- License: MIT
- Architecture: Whisper (GGML)
LA Studio Notes
Whisper.cpp in LA Studio runs inference locally using CPU and GPU options. It downloads GGML-converted models from conversion repository ggerganov/whisper.cpp. It needs:
- Backbone:
ggml-base.bin(or other supported GGML files).
It supports 100 languages, automatic language detection, translation to English, token timestamps, and speaker diarization workflows.
Configure Workflow Parameters
Configure these custom parameters to fine-tune the speech and voice workflows.
Temperaturefloat
Controls randomness of speech generation. Higher values increase variety.
1
Seedint
Random seed for reproducibility. Set to -1 for random.
-1
Max Codec Stepsint
Maximum decoding steps for the neural voice audio synthesizer.
100