CHiME-5 Speech Recognition Dataset
Date
3 years ago
Publish URL
Categories

CHiME-5 is a dataset about automatic speech recognition processing technology. The dataset comes from the fifth CHiME challenge and includes tasks of remote multi-microphone conversations in real home environments. The speech material is extracted from a dinner scene. The dataset obtains natural conversation speech data and is recorded by 6 Kinect microphone arrays and 4 binaural microphone pairs.
The dataset includes data collection procedures, tasks, and baseline systems for single-array and multi-array track language modeling as well as array synchronization, speech enhancement, conventional, and end-to-end ASR.