TTS

This dataset was recorded by 620 speakers with authentic pronunciation and diverse vocal qualities (334 males and 286 females) in a quiet indoor environment. The recorded texts cover all phonemes, and the annotators have a professional linguistic background, ensuring the data meets the research and development needs for voice synthesis.

50 speakers, gender balanced, with pronunciation and prosody annotations

This dataset was recorded by 5 speakers with authentic pronunciation and diverse vocal qualities (one male and four females) in a professional recording studio. The recorded texts span the full range of phonemes, and the annotators have a professional linguistic background, ensuring the data meets the research and development needs for voice synthesis.

This dataset was recorded by a 59-year-old female speaker with authentic pronunciation and a mature, elegant vocal quality in a professional recording studio. The recorded texts cover all phonemes, and the annotators have a professional linguistic background, ensuring the data meets the research and development needs for voice synthesis.

[Emotion] Five emotions: Excited, Sad, Angry, Fearful, and Empathetic. [Duration] Approximately 2 hours per speaker.

This dataset was recorded by a 42-year-old male speaker with authentic pronunciation and a mature, refined vocal quality in a professional recording studio. The recorded texts span the full range of phonemes, and the annotators have a professional linguistic background, ensuring the data meets the research and development needs for voice synthesis.

[Style] Voice Assistant style, Podcast style, Audiobook style, and Online Learning style (four styles in total). [Duration] Approximately 2 hours per speaker, with about 0.5 hours for each style. [Content] Voice assistant data is recorded as single utterances, while the other styles are recorded in paragraph-level segments.