1. Speech Recognition in Noise
Esfandiar Zavarehei
Department of Electronic and Computer Engineering
Brunel University
25 May, 2004

2. Contents The use of formant features in speech recognition
- Variable-Order LP Formant Tracker with Kalman Filtering
- Results
Kalman De-noising
- Tracking and Filtering the Frequency Trajectories (RASTA)
- How Kalman Filter is applied to de-noising problem
- Advantages of Kalman

3. Variable-Order LP Formant tracker
Higher order of LP modelling for higher resolution
Continuity criteria for better classification
Kalman Filtering for smoother Tracks

4. Formant Feature (FF) Vectors
In addition to the Frequency of poles their Band Widths and Magnitudes are used as well
The HMM models are trained on mono-phones.

5. FF vs. MFCC with and without energy component
Mono-phone recognition in Train noise
Better performance of FF in severe noisy conditions

6. Robustness of dynamic FF to noise
Mono-phone recognition in Train noise
Dynamic Features are much more robust to noise

7. The use of the Formants for consonant recognition
Mono-phone recognition in Train noise
Higher Recognition rates than vowels in higher SNR
More sensitive to noise because of the lower energy level

8. De-noising the speech by filtering frequency trajectories

9. RelAtive SpecTrA (RASTA) Processing
Filtering the frequency trajectories of the cubic root of power spectrum using a fixed IIR filter

10. The use of FIR filters in RASTA
Filtering the frequency trajectories of the power spectrum using a bank of non-casual FIR filters
not adaptive
experimentally derived
Filters’ Impulse Response

11. Kalman Filter adaptively updates itself with noise covariance
Kalman Filtering
Kalman Filter adaptively updates itself with noise covariance

12. How Kalman Filter is applied to de-noising problem
Noise Modelling and updating
Neighbour Trajectory
Segment Frequency Bin Trajectory
VAD
Noise Modelling
Prior Noise Model and Trajectory Statistics
Spectral Subtraction
Observation
Predictor
Predicted
Error covariance
Noise Covariance
Mean
Kalman Gain
Estimator
Output
Kalman Filtering

13. A more informed noise reduction
Advantages of Kalman
A more informed noise reduction
Combining the prediction and the observation of the frequency trajectory
Adaptively updating the noise model while filtering the trajectory (in comparison with RASTA)
Could (and probably should) be combined with spectral subtraction for improved performance