ModelsLab/Obama-Sample-Dataset

---
license: mit
language:
- en
metrics:
- wer
pipeline_tag: audio-to-audio
tags:
- rvc
- audio
- speech
---

# Obama Voice Sample Dataset for RVC Training

A curated dataset of Barack Obama's voice samples, specifically prepared for training Demo RVC (Retrieval-based Voice Conversion) model on ModelsLab.

## Dataset Specifications
- Total Duration: 25+ minutes
- Audio Format: WAV
- Sampling Rate: 24 kHz
- Content Type: Clean speech samples from speeches and addresses

## Usage
This dataset is designed for training RVC (Retrieval-based Voice Conversion) models. The minimum recommended training duration for RVC is 25 minutes of clean audio, which this dataset provides.

## Important Notes
- Use this dataset responsibly and in accordance with ethical AI guidelines
- The dataset is intended for research and educational purposes
- Please check local regulations regarding voice cloning and impersonation

## Source
Audio samples are collected from public speeches and addresses given by Barack Obama.

## Requirements for RVC Training
- Minimum audio length: 25 minutes
- Format: WAV files
- Sample rate: 24 kHz (minimum)
- Clean audio without background noise

许可证：MIT
语言：英语
评估指标：词错误率（Word Error Rate，WER）
任务流水线标签：音频到音频
标签：RVC、音频、语音


# 用于RVC（基于检索的语音转换，Retrieval-based Voice Conversion）训练的奥巴马语音样本数据集

本数据集为精选的巴拉克·奥巴马语音样本，专为ModelsLab平台上的Demo RVC模型训练打造。

## 数据集规格
- 总时长：25分钟以上
- 音频格式：WAV
- 采样率：24 kHz
- 内容类型：取自公开演讲与致辞的纯净语音样本

## 使用说明
本数据集专为训练RVC（基于检索的语音转换）模型设计。RVC训练的最小推荐纯净音频时长为25分钟，本数据集恰好满足该要求。

## 重要提示
- 请负责任地使用本数据集，并遵守人工智能伦理准则
- 本数据集仅用于研究与教育用途
- 请核实当地关于语音克隆与声音模仿的相关法规

## 数据来源
本数据集的音频样本取自巴拉克·奥巴马的公开演讲与致辞。

## RVC训练要求
- 最小音频时长：25分钟
- 格式：WAV文件
- 采样率：24 kHz（最低要求）
- 无背景噪音的纯净音频