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). • 233 items • Updated • 2
nvidia-parakeet-rnnt-1.1b-onnx
ASR model exported to ONNX for the onnx-asr library and the ovos-stt-plugin-onnx-asr OpenVoiceOS STT plugin. Ships both fp32 and int8 weights.
Language: English. Converted from nvidia/parakeet-rnnt-1.1b.
import onnx_asr
model = onnx_asr.load_model("OpenVoiceOS/nvidia-parakeet-rnnt-1.1b-onnx")
print(model.recognize("audio.wav"))
- Downloads last month
- 35
Model tree for OpenVoiceOS/nvidia-parakeet-rnnt-1.1b-onnx
Base model
nvidia/parakeet-rnnt-1.1b
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.ASRModel.from_pretrained("OpenVoiceOS/nvidia-parakeet-rnnt-1.1b-onnx") transcriptions = asr_model.transcribe(["file.wav"])