Analysis of Embedded AI Speech Recognition Technology Based on MFCC

With the rapid development of artificial intelligence technology, speech recognition has been widely used as an important human-computer interaction method; and the emergence of embedded AI technology has led to a wider application of speech recognition in end products such as smart speakers, smart phones, smart watches, etc. By performing real-time speech signal processing and model inference on local devices, the speed and accuracy of recognition can be improved, and it can work normally in a network-less environment. With the continuous development of hardware and software, speech recognition embedded AI technology is expected to achieve faster and smarter voice interaction. This thesis analyses the embedded AI speech recognition technology based on MFCC and discusses its technical principles, test reviews and many challenges.