A 75-hour Taiwan Mandarin Speech Recognition Database
A 75-hour Taiwan mandarin corpus for recognition has been released. The speech samples are collected in Taiwan. In total of 100 native Taiwan speakers (1:1 gender distribution) from the major areas of Taiwan are carefully selected.
The corpus has effective duration of 75 hours. It is recorded using 16 kHz sampling frequency and 16-bit quantization accuracy in mono PCM format.
The corpus is collected using various Android smart models and primarily in indoor environment.
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 Taiwan mandarin 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.