Synthetic Speech Database

A synthetic speech database totaling 300,000 hours. The data is primarily sourced from audio platforms such as Himalaya, Yunting, Lizhi FM, audiobooks, and Tencent Music. It covers a wide range of domains, including children’s content, health, current affairs, history, business, daily life, and entertainment. The sampling rate is 16,000 Hz, with a bit depth of 16 bits. This database was collected from publicly available online audio platforms and has undergone technical preprocessing to ensure high-quality audio suitable for speech synthesis. The preprocessing steps are as follows: Data naming conventions:

├─SPEECHDATA (audio folder)
││├─XIMALAYA (source category)
││ ││ XIW00000001.wav (audio file)
││├─DOC
││ ││ content.txt (transcribed text for the audio)
││├─LIZHI

Sample Data Download

Click to download the sample dataset (sample_data.zip)