Model: parakeet-v3-int8
Original model: NVIDIA Corporation, https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3
ONNX export: Ilya Stupakov, https://huggingface.co/istupakov/parakeet-tdt-0.6b-v3-onnx
Revision: 8f23f0c03c8761650bdb5b40aaf3e40d2c15f1ce
Source: https://huggingface.co/istupakov/parakeet-tdt-0.6b-v3-onnx/tree/8f23f0c03c8761650bdb5b40aaf3e40d2c15f1ce
License: Creative Commons Attribution 4.0 International (CC BY 4.0)
https://creativecommons.org/licenses/by/4.0/

This is the upstream ONNX export with int8 quantized weights, selected without
changes to the downloaded files. TBridge supplies a separate speech-model.json
configuration. The model is not covered by the TBridge Perimeter license.
The authors do not endorse TBridge. See the accompanying CC-BY-4.0.txt.

Upstream model card (unchanged):

---
license: cc-by-4.0
language:
  - en
  - es
  - fr
  - de
  - bg
  - hr
  - cs
  - da
  - nl
  - et
  - fi
  - el
  - hu
  - it
  - lv
  - lt
  - mt
  - pl
  - pt
  - ro
  - sk
  - sl
  - sv
  - ru
  - uk
base_model:
- nvidia/parakeet-tdt-0.6b-v3
pipeline_tag: automatic-speech-recognition
tags:
- automatic-speech-recognition
- asr
- onnx
- onnx-asr
---

NVIDIA Parakeet TDT 0.6B V3 (Multilingual) [model](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3) converted to ONNX format for [onnx-asr](https://github.com/istupakov/onnx-asr).

Install onnx-asr
```shell
pip install onnx-asr[cpu,hub]
```

Load Parakeet TDT model and recognize wav file
```py
import onnx_asr
model = onnx_asr.load_model("nemo-parakeet-tdt-0.6b-v3")
print(model.recognize("test.wav"))
```

Code for models export
```py
import nemo.collections.asr as nemo_asr
from pathlib import Path

model = nemo_asr.models.ASRModel.from_pretrained("nvidia/parakeet-tdt-0.6b-v3")

onnx_dir = Path("nemo-onnx")
onnx_dir.mkdir(exist_ok=True)
model.export(str(Path(onnx_dir, "model.onnx")))

with Path(onnx_dir, "vocab.txt").open("wt") as f:
    for i, token in enumerate([*model.tokenizer.vocab, "<blk>"]):
        f.write(f"{token} {i}\n")
```