The Advantage of Implementing Martin’s Noise Reduction Algorithm in Critical Bands Using Wavelet Packet Decomposition and Hilbert Transform
2008
In this paper we address the problem of enhancing single channel speech signal corrupted with additive background noise. We present a new scheme which utilizes a different time frequency representation along with the psychoacoustic features of human ear and combines these features with the well-known noise estimation method of minimum tracking. Instead of Fourier transform, we use a perceptual wavelet packet decomposition of speech, and perform spectral tracking and filtering on the envelope of the analytic signal.
Keywords:
- Constant Q transform
- Discrete wavelet transform
- Stationary wavelet transform
- Harmonic wavelet transform
- Second-generation wavelet transform
- Hilbert spectral analysis
- Electronic engineering
- Wavelet
- Artificial intelligence
- Pattern recognition
- Mathematics
- Wavelet packet decomposition
- Wavelet transform
- Algorithm
- Spectral density estimation
- Speech recognition
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
11
References
2
Citations
NaN
KQI