The role of adaptive filters in the recognition of speech commands

This article discusses adaptive methods for filtering noise affecting speech commands in real operating conditions. The method of adaptive NLMS filters is proposed for speech command recognition. The recognition accuracy of speech command has increased due to the use of the NLMS method. Digital signal filters are used to eliminate interference in speech recognition systems. However, the effectiveness of the use of classical filtering algorithms for speech signals with a complex appearance is very low. In such cases, the use of adaptive filters can meet the system requirement. Flexible filters use Least Mean Square (LMS), Normalized LMS (NLMS) algorithms to eliminate noise. These algorithms differ from each other in Global Signal-to-noise ratio (GSNR). If the GSNR is good, this algorithm is more useful for the system. Using these algorithms, the accuracy of the speech recognition system was found. Thus, we conclude that adaptive filtering algorithms are more efficient than classical filtering algorithms.

Перевод пока недоступен