This dataset was recorded by 25 speakers with authentic pronunciation and a variety of vocal qualities (12 males and 13 females) in a professional recording studio. The recorded texts encompass 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.
Premium Chinese Female Voice Speech Synthesis Corpus (3 Speakers)
Features: Agent-style duplex conversation; the main speaker acts as the Agent and engages in multi-turn dialogues with companion voices in scenarios such as emotional companionship and daily Q&A.
Emotion: Both main and companion voices are annotated with 28 emotions—including neutral, joy, anger, sadness, fear, surprise, disgust, etc.—with clause-level emotion tagging.
Standard Arabic Male Voice Speech Synthesis Corpus
Covers news, dialogues, encyclopedic content, time, date, menus, hotels, and English abbreviations, suitable for multi-scenario, multi-domain speech synthesis.
