1. Presented By: Shamil. C Roll no: 68 E.I Guided By: Asif Ali Lecturer in E.I

2.  Introduction  Speech measurement with LDV  Principe of LDV  Measurement Setup  Problem formulation  Speech Enhancement Algorithm  Speckle noise suppression  LDV-Based time frequency VAD  Spectral gain modification  Experimental Results  Conclusion

3.  Achieving high speech intelligibility in noisy environments is one of the most challenging and important problems for existing speech- enhancement and speech-recognition systems.  Recently, several approaches have been proposed that make use of auxiliary non acoustic sensors, such as bone and throat- microphones.  Major drawback of most existing sensors is the requirement for a physical contact between the sensor and the speaker.  Here present an alternative approach that enables a remote measurement of speech, using an auxiliary laser Doppler vibrometer (LDV) sensor.

5.  fd(t) = 2ν(t) cos(α)/λ ν(t)=> instantaneous throat-vibrational velocity α => Angle between the object beam and the velocity vector λ =>laser wavelength.  LDV-output signal after an FM-demodulator is Z(t) = f b + [2Av cos(α)/λ].cos(2πf v t). (1)

7.  Employing the VibroMet™500V LDV.  Consists of a remote laser-sensor head and an electronic controller.  Operates at 780 nm wavelength.  Can detect vibration frequencies from DC to over 40 kHz.  Its operational working distance ranges from 1 cm to 5 m.

9.  let y(n) =x(n) + d(n) y(n)-observed signal in the acoustic sensor. x(n) -Speech signal. d(n)-Un correlated additive noise signal.  In the STFT domain, Y lk = X lk + D lk Where l= 0, 1,... is the frame index. k = 0, 1,..., N − 1is the frequency- bin index.

10. Use overlapping frames of N samples with a framing-step of M samples. Let H 0lk and H 1lk indicate, respectively, speech absence and presence hypotheses in the time-frequency bin (l, k), i.e., H 0lk : Y lk = D lk H 1lk : Y lk = X lk + D lk. X̂ lk = G lk Y lk.

11.  The OM-LSA estimator minimizes the log spectral amplitude under signal presence uncertainty resulting in, G lk = {G H1lk }ˆP lk.G min ˆ1 − P lk. Where, G H1lk is a conditional gain function given H 1lk & G min << 1 is a constant attenuation factor. P lk is the conditional speech presence probability.

12.  Denoting by ξlk and γlk we get, is the a priori probability for speech absence, -Posteriori SNR -Priori SNR

13.  Speckle-Noise Suppression The output of the speckle-noise detector is, W l (n) = G l Z l (n) Where G l = Gs min <<1 for I l = 1(speckle noise is present) G l = 1 otherwise.

15. -Represents the noise-estimate bias -Smoothed-version of the power spectrum Then, we propose the following soft- decision VAD:

17. Speech in a given frame is defined by We attenuate high-energy transient components to the level of the stationary background noise by updating the gain floor to -Stationary noise-spectrum estimate -Smoothed noisy spectrum

19.  Speckle noise was successfully attenuated from the LDV-measured signal using a kurtosis-based decision rule.  A soft-decision VAD was derived in the time-frequency domain and the gain function of the OM-LSA algorithm was appropriately modified.  The effectiveness of the proposed approach in suppressing highly non-stationary noise components was demonstrated.

20.  I. Cohen and B. Berdugo, “Speech enhancement for nonstationary noise environment,” Signal Process., vol. 81  T. F. Quatieri, K. Brady, D. Messing, J. P. Campbell, W. M. Campbell, M. S. Brandstein, C. J.Weinstein, J. D. Tardelli, and P. D. Gatewood, “Exploiting nonacoustic sensors for speech encoding,”  T. Dekens, W. Verhelst, F. Capman, and F. Beaugendre, “Improved speech recognition in noisy environments by using a throat microphone for accurate voicing detection,” in 18th European Signal Processing Conf. (EUSIPCO), Aallborg, Denmark, Aug. 2010, pp. 23–27  M. Johansmann, G. Siegmund, and M. Pineda, “Targeting the limits of laser doppler vibrometry,”  http://www.metrolaserinc.com