Views
No views yet
pip install onnx-asr[cpu,hub]1import onnx_asr
2model = onnx_asr.load_model("nemo-parakeet-tdt-0.6b-v2")
3print(model.recognize("test.wav"))1import nemo.collections.asr as nemo_asr
2from pathlib import Path
3
4model = nemo_asr.models.ASRModel.from_pretrained("nvidia/parakeet-tdt-0.6b-v2")
5
6onnx_dir = Path("nemo-onnx")
7onnx_dir.mkdir(exist_ok=True)
8model.export(str(Path(onnx_dir, "model.onnx")))
9
10with Path(onnx_dir, "vocab.txt").open("wt") as f:
11 for i, token in enumerate([*model.tokenizer.vocab, "<blk>"]):
12 f.write(f"{token} {i}\n")