Audio Activity Detection for Watermarking Based on Wavelet Denoising

Objectives: The watermark embedding in Non-Activity Region (NAR) of the underlying audio signal results for the degradation of audio quality. This research focused on the development of audio activity detection method for the audio watermarking procedure to prevent from watermark embedding in the NAR. Methods/Statistical Analysis: We propose the audio activity detection method based on the wavelet denoising algorithm to classify the NAR in audio signal. We experiment and apply three- type wavelet denoising algorithms, subsequently the wavelet denoising algorithms showed the same performance. We analyze the audio quality by the Signal to Noise Ratio (SNR) for two cases that the embedding and non-embedding of watermark in the NAR. Findings: We surveyed the effects on the watermarking embedding in the NAR. Decreasing a payload on watermarking, the watermark embedding in the NAR increases the audio quality in SNR and audio test by MUSHRA (Multi Stimulus Test with Hidden Reference and Anchor). In extraction process the extraction rate of embedded watermark is slightly increasing when our method is applied. In comparisons of the other research results such as DFT (Discrete Fourier Transform) based method and MFCC (Mel Frequency Cepstral Coefficients) based method, our method has the better accuracy in AAD. And also our method is in low complexity for the analysis of computational complexity by the effective computational structure of filter bank for implementation of wavelet filter. In additional research results, we find out that the audio signal in the NAR has a random- like pattern by the analysis of power spectrum of it.In statistical experiment a signal in NAR analyzed by statistical parameters such as mean and variance. As a result, our method is effective for computational complexity and produces good audio quality for the watermark embedding procedure. Improvements/Applications: Our method for audio activity detection is very effective computational complexity and produces high audio quality for watermark embedding procedure. It is applicable for the audio watermarking system for high- quality watermarked audio signal.