DSPP APPLICATION

The last experiment, was a group experiment. We were asked to perform signal processing on a 1D signal and find out an application for the same.  The application that we selected was 'Noise Reduction Using Adaptive Filters'. The group members are Kapil Rawal, Prerana Sarode,Chinmay Upadhye and Harshit Shukla.

Summary of the paper I referred to.

A new adaptive speech noise removal algorithm was proposed based on a two-stage Wiener filtering. A first Wiener filter is used to produce a smoothed estimate of the a priori signal-to-noise ratio (SNR), aided by a classifier that separates speech from noise frames, and a second Wiener filter is used to generate the final output. Spectral analysis and synthesis is performed by a modulated complex lapped transform (MCLT).

Patent

An array of microphones utilizes two sets of two microphones for noise suppression. A primary microphone and secondary microphone of the three microphones may be positioned closely spaced to each other to provide acoustic signals used to achieve noise cancellation. A tertiary microphone may be spaced with respect to either the primary microphone or the secondary microphone in a spread-microphone configuration for deriving level cues from audio signals provided by tertiary and the primary or secondary microphone.

https://drive.google.com/open?id=0B8F3pY6H1pIWcDZoalBRWnR1M2M
https://drive.google.com/open?id=0B8F3pY6H1pIWajRuRHo4Q1hYUW8
https://drive.google.com/open?id=0B8F3pY6H1pIWTHB0bzFWV2lBbVE
https://drive.google.com/open?id=0B8F3pY6H1pIWaDZVV2QtamF4SEk

http://preranasarode1995.blogspot.com/2016/04/the-last-experiment-was-group-experiment.html

DSP Processor

Experiment 9 of the course was the first hardware based demo experiment. One of our seniors showed us the demo on operations such as additionubtraction multiplication and Logical Operations and Shifting operations. It was concluded that a large number of mathematical operations can be performed and repeatedly on this processor since it is a specialised microprocessor.

http://preranasarode1995.blogspot.com/2016/04/dsp-processor.html

FIR Filter Design Using Frequency Sampling

During this lab session we needed to perform FIR using Frequency sampling for LPF and HPF and BPF. It was implemented in Scilab.As filter order increases, number of lobes in stop band also increases. The output is also symmetric in nature. 

In this the desired frequency response Hd(w) is sampled at w=(2*pi*k)/N and the frequency samples thus obtained are taken as DFT coefficients. FIR filter with impulse response is then calculated by IDFT.

We observed that the stop band attenuation values computed in the code are close to the values taken as input. We also looked at the magnitude plots and observed the presence of lobes in the stop band for low pass filter .

https://drive.google.com/open?id=0B8F3pY6H1pIWRHVRbnBZMnVqUDg

http://preranasarode1995.blogspot.com/2016/04/fir-filter-design-using-frequency.html

FIR Filter Design Using Window Method

 The user was prompted to input values like Attenuation in Stop band (As) and Pass band (Ap) as well as Pass band frequency, Stop band frequency and sampling frequency.A low pass and Band pass filter was designed.

The program calculated which windowing function would be most suitable for the given input specifications.Thus, we performed the experiment for two window functions: Bartlett and Hanning.

 The magnitude and phase plot of both the filters was plotted using scilab. In this method, the desired impulse response is multiplied with window function w(n) to obtain h(n) which after Z-transfrom  gave H(z). The phase  plot being linear, there will be no distortion at the output.

We learnt that Hamming Window function gives more attenuation in stop band than Hanning window, hence it is a better window function.

https://drive.google.com/open?id=0B8F3pY6H1pIWM2poeTlLX2ZxVjg
https://drive.google.com/open?id=0B8F3pY6H1pIWMzhxRy1nSlE0S1U
http://preranasarode1995.blogspot.com/2016/04/fir-filter-design-using-window-method.html

Chebyshev Filter Design

Th filter was designed just the way it was discussed in the class. The input parameters were As Ap Pass Band the input parameters As ,Ap pass band frequency and stop band frequency and sampling frequency.The pole zero plot was also drawn and the values of As and Ap were compared from the magnitude spectrum. The poles lie within the unit cirle. The number of ripples representated the order of the filter.  We observed the following things :

1. In the magnitude response of the Low Pass Filter there were ripples in the pass band whereas the stop band is monotonic and there is no ripple.

2. In the magnitude response of the High Pass, there were ripple in the pass band whereas the response is monotonic in the stop band.

https://drive.google.com/open?id=0B8F3pY6H1pIWejYtU3NzUUVqOFE

http://preranasarode1995.blogspot.com/2016/04/chebyshev-filter-design.html

Butterworth Filter Design

The butterworth filter design was done for Low Pass and High Pass. It was done with specific input parameters like As, Ap , Pass Band frequency, Stop Band Frequency and Sampling frequency. H(z) for each were calculated and so was the order of the filter. The theoretical and observed values were compared. Maginitude spectrum and pole zero plot was drawn and it was observed that butterworth filters are monotonic in its pass band and stop band.

It was noticed that the magnitude response smooth in both pass band and stop band. With higher order, the magnitude response became sharoer and started resembling the ideal filter more closely.

http://preranasarode1995.blogspot.com/2016/04/butterworth-filter-design.html


https://drive.google.com/open?id=0B8F3pY6H1pIWanN3aHlQbGlCRWM
https://drive.google.com/open?id=0B8F3pY6H1pIWMGliNXcxd1NtWnM

OAM and OSM

Linear Convolution was performed using Overlap Add method  for which the length of the ip signal x(n) was 12 and h(n) was 3. The program was done by breaking the ip signal in 2 parts of length 6 each and the length of the decomposed signal was 8. So two outputs were generated and the final output was stored in y(n) of length 14. Overlap Save Method involved the same procedure and in the earlier case x1(n) and x2(n) but in OSM x1(n) s2(n) and x3(n) were generated. It was noted that OAM and OSM can be used to filter long ip sequences using FFT.

We realised that it had an important limitation wherein it required all input values to be available for giving an output. For a real world signal which can be arbitrarily long, waiting for the entire signal to arrive and get stored would cause massive delays and also increase the cost of storage equipment.

https://drive.google.com/open?id=0B8F3pY6H1pIWaGlGbVgwX2YzMUE
https://drive.google.com/open?id=0B8F3pY6H1pIWWVE5Y0xJOUhoZzg

http://preranasarode1995.blogspot.com/2016/04/oam-and-osm.html