README.md

---
language:
- it
license: apache-2.0
tags:
- text-to-speech
- tts
- italian
- qwen
- audio
pipeline_tag: text-to-speech
---

# Italian TTS Model - Qwen3-TTS Fine-tuned

This is an Italian Text-to-Speech model fine-tuned from Qwen3-TTS-12Hz-1.7B-Base.

## Model Details

- **Base Model:** Qwen3-TTS-12Hz-1.7B-Base
- **Language:** Italian (Italiano)
- **Training Data:** 115,000 Italian audio samples
- **Fine-tuning Parameters:**
  - Batch size: 8
  - Learning rate: 1e-5
  - Epochs: 10
  - Gradient accumulation: 4
  - Mixed precision: bf16

## Usage

```python
import torch
from qwen_tts.inference.qwen3_tts_model import Qwen3TTSModel

# Load model
model = Qwen3TTSModel.from_pretrained(
    "Aynursusuz/Qwen-TTS-Best-Model",
    torch_dtype=torch.bfloat16,
)

# Generate speech
text = "Buongiorno, come stai oggi?"
audio = model.inference(
    text=text,
    speaker="italian_voice"
)
```

## Training Details

This model was trained on a high-quality Italian speech dataset with the following configuration:
- GPU: NVIDIA A100 80GB
- Training time: ~15-20 hours
- Optimizer: AdamW with weight decay 0.01
- Best checkpoint selected based on validation loss

## Citation

```bibtex
@misc{qwen-tts-italian-2026,
  author = {Aynur},
  title = {Italian TTS Model based on Qwen3-TTS},
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/Aynursusuz/Qwen-TTS-Best-Model}}
}
```

## License

This model inherits the Apache 2.0 license from the base Qwen3-TTS model.