STT/ASR - onnx
Collection
OVOS STT/ASR models for onnx-asr (ONNX runtime). Most ship fp32 + int8 (set quantization: int8 for faster/smaller CPU inference). • 234 items • Updated • 2
CAMOES-whisper-asr — ONNX
ONNX export of amalia-llm/CAMOES-whisper-asr (Whisper large-v3 fine-tuned for
European Portuguese, CAMOES project) for onnx-asr
(standard whisper model type — works with stock onnx-asr, no patches needed).
fp32 and int8 variants included; int8 decoder produced by quantizing the pre-merge
decoders and re-merging (the merged graph's If-subgraphs defeat direct quantization).
Usage
import onnx_asr
model = onnx_asr.load_model("whisper", "path/to/this/repo") # or quantization="int8"
print(model.recognize("audio_16khz.wav", language="pt"))
Verified on FLEURS Portuguese clips (fluent EP output, fp32 and int8). RTF on an AMD Ryzen 5 7600 (CPU): ~0.8 fp32, ~0.4 int8.
- Downloads last month
- 2,331