1 / 18

Reduction of Additive Noise in the Digital Processing of Speech

Reduction of Additive Noise in the Digital Processing of Speech. Avner Halevy AMSC 663 Mid Year Progress Report December 2008 Professor Radu Balan. Brief Reminder. Goal: reduce additive white Gaussian noise degrading a speech signal Use short time analysis (frames) Algorithms:

hagop
Télécharger la présentation

Reduction of Additive Noise in the Digital Processing of Speech

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Reduction of Additive Noise in the Digital Processing of Speech Avner Halevy AMSC 663 Mid Year Progress Report December 2008 Professor RaduBalan

  2. Brief Reminder • Goal: reduce additive white Gaussian noise degrading a speech signal • Use short time analysis (frames) • Algorithms: • Spectral subtraction (done) • Iterative Wiener filtering (next) • Test objectively in lab conditions

  3. Spectral Subtraction • Estimate the magnitude of the noise spectrum when speech is absent • Subtract the estimate from the magnitude of the spectrum of the noisy signal • Keep the noisy phase • Inverse Fourier to obtain enhanced signal in time domain • Underlying statistical assumptions

  4. Spectral Subtraction

  5. SS – “Filtering” View

  6. Spectral Subtraction Implemented • Implementation issues • Choice of analysis and synthesis windows • Ensuring nonnegative magnitude • Estimation of noise spectrum • Choice of exponent • Validation • Variations

  7. Choice of Windows

  8. SS – Preliminary Results

  9. SS – Preliminary Results Clean Frequency (Hz) Time (sec)

  10. SS – Preliminary Results Clean Noisy

  11. SS – Preliminary Results

  12. SS – Preliminary Results SS – Preliminary Results Noisy Enhanced

  13. Evaluation of Exponent Values tested: p = .1, .5, 1, 1.5, 2, 3, 4, 5

  14. Testing • TIMIT – database of 6300 sentences for evaluation of speech processing algorithms • Segmental SNR • Segmental “Filtering” SNR • Segmental “Filtering” distortion • Future: Perceptual Evaluation of Speech Quality (PESQ) – telecomm standard

  15. Shortcomings of SS • Using the noisy phase (in low SNR) • “musical noise” artifacts caused by • Error in noise spectrum estimation • Flooring of negative components • Fluctuations in signal spectrum • Solutions: • Over subtraction • Smoothing of signal spectrum

  16. Dealing with “Musical Noise” Over subtraction Smoothing signal spectrum

  17. Next Steps • Test more signals • Evaluate using PESQ • Implement Wiener filtering • Further testing • Compare performance

  18. Bibliography • [1] Deller, J., Hansen, J., and Proakis, J. (2000) Discrete Time Processing of Speech Signals, New York, NY: Institute of Electrical and Electronics Engineers • [2] Quatieri, T. (2002) Discrete Time Speech Signal Processing, Upper Saddle River, NJ: Prentice Hall • [3] Loizou, P. (2007) Speech Enhancement: Theory and Practice, Boca Raton, FL: Taylor & Francis Group • [4] Rabiner, L., Schafer, R. (1978) Digital Processing of Speech Signals, Englewood Cliffs, NJ: Prentice Hall

More Related