Litcius/Paper detail

Adaptive Algorithms for Signature Wavelet recognition in the Musical Sounds

Duraipandian M.

2020Journal of Soft Computing Paradigm35 citationsDOIOpen Access PDF

Abstract

The scaling and as well as the wavelet-functions of the wavelet is detected engaging the wavelet-filters that are empowered with the filter-bank principle that is utilized in recognizing the rough calculation and the feature co-efficient of the wavelet-filter. The coefficients recognized by the filter-bank for the musical sounds produced by the musical-instruments enables one to have a signature-wavelet of the sound signal. The signature-wavelet renovates the actual musical signal with insignificant disturbance. In order to recognize the factors (coefficients) the paper employs the least mean square (L-MS), normalized least means square (NL-MS), recursive least square (R-LS) and the QR-Recursive least square (QR-RLS). Among the above four the R-LS and the QR-RLS performs well under all grounds. More over the algorithm converges swiftly compared to the other algorithm. Thus providing an accuracy and SOC (speed of convergence) improved scaling and wavelet-function recognition.

Topics & Concepts

WaveletAlgorithmMathematicsFilter (signal processing)Cascade algorithmSpeech recognitionPattern recognition (psychology)Signature (topology)Recursive least squares filterFeature (linguistics)Filter bankComputer scienceLeast mean squares filterAdaptive filterSquare (algebra)Wavelet transformWavelet packet decompositionArtificial intelligenceComputer visionPhilosophyLinguisticsGeometryImage and Signal Denoising MethodsBlind Source Separation TechniquesSpeech and Audio Processing