A 1500-Hour American English Speech Recognition Database
A 1500-hour American English Corpus for Recognition has been released. It is recorded in real environments using 16 kHz sampling frequency and 16-bit quantization accuracy mono PCM format with smart phone and 48 kHz sampling frequency and 16-bit quantization accuracy mono PCM format with desktop microphone.
1400 native American English speakers (1:1 gender distribution) with their respected dialectal characteristics are carefully selected.
This corpus can be used for training and testing the speech recognition system, as well as speech analysis. It has been well-acknowledged by industry as a corpus with high speech quality and recognition accuracy.