erminga/emo-tts

Emo-TTS评估数据集是一个用于评估高唤醒情感语音合成的数据集集合，包含HIED、ESD、EmoV-DB和Expresso四个子数据集。HIED是一个专门设计用于测试高唤醒情感条件下TTS系统的评估基准，包含400个样本，涵盖愤怒、快乐、悲伤和惊讶四种情感。ESD是一个情感语音数据集，包含中性和四种基本情感（快乐、悲伤、愤怒、惊讶）的语音样本，由10名英语和10名汉语说话者录制。EmoV-DB是一个情感语音数据库，包含中性、愉快、愤怒、困倦和厌恶五种情感，由4名说话者录制。Expresso是一个高质量的富有表现力的语音数据集，包含8种朗读风格和26种即兴对话风格，由4名说话者录制。这些数据集支持多种语言，包括英语、汉语和法语，并提供了详细的元数据和音频波形。

The Emo-TTS Evaluation Datasets is a collection of datasets used for evaluating high-arousal emotional speech synthesis, comprising four sub-datasets: HIED, ESD, EmoV-DB, and Expresso. HIED is a benchmark specifically designed to test TTS systems under high-arousal emotional conditions, containing 400 samples covering four emotions: angry, happy, sad, and surprise. ESD is an emotional speech dataset with neutral and four basic emotions (happy, sad, angry, surprise), recorded by 10 English and 10 Chinese speakers. EmoV-DB is an emotional voices database with five emotions: neutral, amused, angry, sleepy, and disgusted, recorded by 4 speakers. Expresso is a high-quality expressive speech dataset with 8 read speech styles and 26 improvised dialogue styles, recorded by 4 speakers. These datasets support multiple languages, including English, Chinese, and French, and provide detailed metadata and audio waveforms.