Active voice detection using ridgelet transform

Abstract EN

Regularly, voice activity detection is the main crucial parameter in any control system using command of voice speech, spatially in environments with noise.

The uses voice activity detection method affects both, complexity of computational and the overall performance of any control system using command of voice speech.

This paper handles the problem of automatic word boundary detection in both cases (the noiseless and noisy backgrounds), by proposed voice activity detection of speech signals based on using Ridgelet Transform (RT).

It uses the Inverse Ridgelet Transform (Ridgelet Transform multiply Ridgelet Transform), which gives the accuracy in environments with noise.

The experimental results point up the effectiveness of the used method in low signal to noise ratio environments, for command voice speech signal detection capability.