README.md

---
license: cc-by-nc-sa-4.0
base_model: macminix/qwen3_voice_design_t1
pipeline_tag: text-to-speech
library_name: transformers
language:
  - en
tags:
  - tts
  - qwen
  - qwen3
  - qwen3-tts
  - voice-design
  - merged-lora
  - fine-tuned
  - audio
---

# ckpt-600 — merged

Self-contained snapshot of [`macminix/qwen3_voice_design_t1`](https://huggingface.co/macminix/qwen3_voice_design_t1)
with the **ckpt-600** LoRA adapter folded into the Talker weights.
No PEFT layers at runtime — load directly with `Qwen3TTSModel.from_pretrained`.

## Quick start

```python
from qwen_tts import Qwen3TTSModel

wrap = Qwen3TTSModel.from_pretrained("<this-repo-id>")
wavs, sr = wrap.generate_voice_design(
    text="Hello, this is a test.",
    instruct="A young adult female speaker speaks calmly at a normal pace.",
    language="english",
    temperature=0.9, top_p=1.0, top_k=50,
    repetition_penalty=1.05, max_new_tokens=600,
)
```

## Repository layout

Same shape as the upstream macminix repo:

- `config.json`, `model.safetensors` — Qwen3-TTS Talker + Code Predictor (merged)
- `speech_tokenizer/` — 12.5 fps × 16 codebook neural codec (unchanged)
- `tokenizer.*`, `vocab.json`, `merges.txt`, `added_tokens.json`,
  `special_tokens_map.json` — Qwen2 BPE tokenizer
- `generation_config.json`, `preprocessor_config.json`
- `vocence_config.yaml`, `chute_config.yml` — runtime + Chutes deploy hints

You can drop your own `miner.py` into this repo (same contract as macminix's:
class `Miner` with `__init__(path_hf_repo: Path)`, `warmup()`,
`generate_wav(instruction, text)` → `(np.ndarray, int)`); the standard
Vocence chute wrapper will load this model unchanged.

## Provenance

See [`merge_info.json`](merge_info.json) for the exact base path, adapter
path, LoRA hyperparameters, and merge timestamp.