---
license: apache-2.0
library_name: transformers
tags:
  - cosyvoice
  - onnx
  - pytorch
  - trust-remote-code
base_model: FunAudioLLM/Fun-CosyVoice3-0.5B-2512
---

# CosyVoice3 Speech Tokenizer PT Parameter Bundle

This repository stores a PyTorch checkpoint converted from ONNX initializers.

## Source

- Base/source model: [FunAudioLLM/Fun-CosyVoice3-0.5B-2512](https://huggingface.co/FunAudioLLM/Fun-CosyVoice3-0.5B-2512)
- Conversion method: ONNX initializer tensors -> `pytorch_model.bin`

## Important Notes

- This repository supports `trust_remote_code=True` loading.
- This is a parameter-store style model bundle, not a full original training architecture reconstruction.
- `configuration_onnx_parameter_store.py`, `modeling_onnx_parameter_store.py`, `config.json`, `onnx_parameter_map.json` are required for `AutoModel.from_pretrained(..., trust_remote_code=True)`.

## Load Example

```python
from transformers import AutoModel

model = AutoModel.from_pretrained(
    "wookee3/cosyvoice3-speech-tokenizer-pt",
    trust_remote_code=True,
)
print(len(model.weights))
```

## Tokenize via Loaded HF Model

```python
from transformers import AutoModel

model = AutoModel.from_pretrained(
    "wookee3/cosyvoice3-speech-tokenizer-pt",
    trust_remote_code=True,
)
# downloads speech_tokenizer_v3.onnx from source repo automatically
indices = model.tokenize_from_file("/path/to/audio.wav")
print(indices.shape)
print(indices[0].tolist()[:30])
```

## Audio -> Token Sequence Example (ONNX Runtime)

This repo does not include ONNX file intentionally. The example script downloads
`speech_tokenizer_v3.onnx` from source repo at runtime.

```bash
pip install numpy soundfile onnxruntime huggingface_hub
python audio_to_tokens_example.py --audio /path/to/audio.wav
```

Or specify a local ONNX path:

```bash
python audio_to_tokens_example.py \
  --audio /path/to/audio.wav \
  --onnx /path/to/speech_tokenizer_v3.onnx
```