xincan/Llama-VITS_data

---
license: mit
dataset_info:
features:
- name: version
dtype: string
- name: data
list:
- name: a
dtype: int64
- name: b
dtype: float64
- name: c
dtype: string
- name: d
dtype: bool
splits:
- name: train
num_bytes: 58
num_examples: 1
download_size: 2749
dataset_size: 58
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
task_categories:
- text-to-speech
language:
- en
---
# Dataset Card for Llama-VITS_data

The dataset repository contains data related with our work "Llama-VITS: Enhancing TTS Synthesis with Semantic Awareness", encapsulating:
- Filtered dataset `EmoV_DB_bea_sem`
- Filelists with semantic embeddings
- Model checkpoints
- Human evaluation templates

## Dataset Details

- **Paper:** Llama-VITS: Enhancing TTS Synthesis with Semantic Awareness
- **Curated by:** Xincan Feng, Akifumi Yoshimoto
- **Funded by:** CyberAgent Inc
- **Repository:** https://github.com/xincanfeng/vitsGPT
- **Demo:** https://xincanfeng.github.io/Llama-VITS_demo/

## Dataset Creation

We fileterd `EmoV_DB_bea_sem` dataset from `EmoV_DB` (Adigwe et al., 2018), a database of emotional speech containing data for male and female actors in English and French. EmoV_DB covers 5 emotion classes, amused, angry, disgusted, neutral, and sleepy. To factor out the effect of different speakers, we filtered the original EmoV_DB dataset into the speech of a specific female English speaker, bea. Then we use Llama2 to predict the emotion label of the transcript chosen from the above 5 emotion classes, and select the audio samples which has the same predicted emotion.
The filtered dataset contains 22.8-minute records for training. We named the filtered dataset `EmoV_DB_bea_sem` and investigated how the semantic embeddings from Llama2 behave in naturalness and expressiveness on it. Please refer to our paper for more information.

## Citation

If our work is useful to you, please cite our paper: "Llama-VITS: Enhancing TTS Synthesis with Semantic Awareness".
```sh
@misc{feng2024llamavits,
title={Llama-VITS: Enhancing TTS Synthesis with Semantic Awareness},
author={Xincan Feng and Akifumi Yoshimoto},
year={2024},
eprint={2404.06714},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```

原始信息汇总

数据集概述

数据集基本信息

• 许可证: MIT
• 数据集大小:
  • 下载大小: 2749字节
  • 数据集大小: 58字节

数据集特征

• 版本 (version): 字符串类型
• 数据 (data):
  • a: 整数类型 (int64)
  • b: 浮点类型 (float64)
  • c: 字符串类型
  • d: 布尔类型 (bool)

数据集分割

• 训练集 (train):
  • 字节数: 58
  • 示例数: 1

任务类别

• 文本到语音 (text-to-speech)

语言

• 英语 (en)