Nexdata/Chinese_Mandarin_Entertainment_anchor_Style_Multi-emotional_Synthesis_Corpus
收藏资源简介：
12小时的中文普通话娱乐主播风格多情感合成语料库，由中文母语者录制，涵盖了六种情感文本和语气词，音素和声调平衡，并由专业语音学家参与标注。数据集格式为48,000Hz, 24bit, 未压缩的wav文件，单声道，录制环境为专业录音室，内容包括七种情感（快乐、愤怒、悲伤、惊讶、恐惧、厌恶）和带有填充词的句子。数据集的语言为普通话，标注包括词和拼音转录、韵律边界标注和音素边界标注。数据量方面，中性数据不少于1.6小时，带有填充词的数据不少于0.4小时，其余六种情感数据每种不少于1.67小时。数据集采用商业许可。
A 12-hour Mandarin Chinese entertainment-host-style multi-emotional speech synthesis corpus recorded by native Chinese speakers. The corpus covers six types of emotional texts and sentence-final particles, with balanced phonemes and tones, and was annotated by professional phoneticians. The dataset is stored as uncompressed single-channel WAV files with a sampling rate of 48,000 Hz and a bit depth of 24 bits. Recorded in professional recording studios, the corpus includes seven emotional categories: joy, anger, sadness, surprise, fear, disgust, and neutral speech, in addition to sentences containing filler words. The official language of the corpus is Mandarin, and the annotations include word and pinyin transcriptions, prosodic boundary annotations, and phoneme boundary annotations. Regarding data volume: the neutral speech data is no less than 1.6 hours, the filler word-containing data is no less than 0.4 hours, and each of the remaining six emotional categories has no less than 1.67 hours of data. The dataset is available under a commercial license.
数据集概述
基本信息
- 名称: Nexdata/Chinese_Mandarin_Entertainment_anchor_Style_Multi-emotional_Synthesis_Corpus
- 语言: 中文（Mandarin）
- 任务类别: 文本到语音（text-to-speech）
数据集描述
- 时长: 12小时
- 内容: 中文普通话娱乐主播风格的多情感合成语料库，由中文母语者录制。包含六种情感文本加语气词，音素和声调平衡。有专业语音学家参与标注。
- 录制环境: 专业录音室
- 录制内容: 包含七种情感（快乐、愤怒、悲伤、惊讶、恐惧、厌恶）及包含填充词的句子
- 演讲者: 专业CharacterVoice，角色为18岁女娱乐主播，喜欢唱歌和跳舞
- 设备: 麦克风
- 标注: 包含字词和拼音转录，韵律边界标注，音素边界标注
数据量
- 中性数据不少于1.6小时
- 包含填充词的数据不少于0.4小时
- 其余六种情感数据每种不少于1.67小时
格式
- 采样率: 48,000Hz
- 位深度: 24bit
- 格式: 未压缩wav，单声道
许可信息
- 许可类型: 商业许可
