A 50-hour English Far Field Command Speech Recognition Database
Far field command word corpus for recognition has been released. 100 male and 100 female speakers from 25 different districts with the age from 17 to 32 are included in the corpus. 200 different English sentences (5 to 50 words) are collected for each talker.
The corpus is collected using various Android smart phone models and primarily in indoor (quiet and anechoic) environments. 6 channels of speech signals are simultaneously recorded, where detailed information is listed below:
Recording Devices |
Audio Format |
Comments |
Multichannel recorder with professional microphones |
48kHz,16bit,4 channel,PCM wave |
one microphone is located close to the talker’s mouth; the other three microphone channels formed an three-element microphone array located at 50 cm from talker’s in the center front position |
Smart Phone Model A |
16kHz, 16bit, stereo, PCM wave |
|
Smart Phone Model B |
16kHz, 16bit, mono, PCM wave |
|
During the corpus design and collection process, we have formed an efficient team with high standard capabilities. This leads to the great success for our corpus to have below 2% sentence-wise error, which is dominating in the current market.
This corpus can be used for training and testing the children 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.