GTCOM's intelligent speech technology platform employs mature multilingual speech recognition and speech synthesis technologies. It can provide speech interaction solutions in Chinese, English, Japanese, Korean, German and Portuguese for real-life speech interaction scenarios.
Speech is the most natural, convenient way for man-machine interaction, and it is an important entrance for the next-generation man-machine interaction. GTCOM is actively pursuing layout in the field of speech. It has developed an internationally advanced speech recognition technology based on based on a self-adaptive, context-dependent, deep-neural-network hidden Markov model. It has also applied deep convolutional neural network technology to acoustic modeling, combined with the end-to-end speech recognition technology based on long short term memory (LSTM) and connectionist temporal classification (CTC), in order to reduce the error rate of recognition and greatly improve the performance of speech recognition.
Based on large-scale and in-depth integration of high-quality corpora, GTCOM has built speech recognition engines covering Chinese, English, Japanese, Korean, German and Portuguese using its independently developed deep learning technology framework, and each language has been supported with more than 10,000 hours of corpora training. Its speech recognition level maintains a leading position in the world. These accurate speech recognition engines have been successfully used in GTCOM's language technology products such as JoveTrans, FindYee App, and YeeCloud Cloud Input Method, laying a good foundation in the speech recognition market. In the future, GTCOM will continue to expand languages and provide speech recognition solutions for users in different languages and on different occasions around the world.