# SynthVoice: This should be a paper Title

📑 [Paper](https://huggingface.co/papers/xxxx.xxxxx)    |    🌐 [Project Page](https://synthvoice.github.io/)    |    💾 [Released Resources](https://huggingface.co/collections/toolevalxm/synthvoice-67a978e28fd926b56a4f55a2)    |    📦 [Repo](https://github.com/xmhtoolathlon/Annoy-DataSync)

This is the resource page of our SynthVoice collection on Huggingface.

**Dataset**

| Dataset | Link |
|-|-|
| SynthVoice-Processed | [🤗](https://huggingface.co/datasets/toolevalxm/SynthVoice-Processed) |

Please also check the raw data: [toolevalxm/SynthVoice-Raw](https://huggingface.co/datasets/toolevalxm/SynthVoice-Raw).

**Models**

| Base Model / Training | SynthVoice | SynthVoice++ |
|-|-|-|
| Coqui TTS VITS | [🤗](https://huggingface.co/toolevalxm/synthvoice-vits) | [🤗](https://huggingface.co/toolevalxm/synthvoice-vits-pp) |

**Introduction**

We utilize the Coqui TTS framework for synthesizing high-quality voice outputs from text transcripts. The synthesis is performed using the VITS model architecture, which has demonstrated superior quality in text-to-speech generation tasks. Our approach involves:

1. Processing raw LibriSpeech transcripts
2. Using Coqui TTS (coqui-ai/TTS) for voice synthesis
3. Post-processing and quality filtering

*Due to licensing requirements, we only release the processed subset containing synthesized outputs.

**License**

The license for this dataset is CC BY 4.0.