Download presentation
Presentation is loading. Please wait.
Published byJonah Logan Modified over 9 years ago
1
Automatic Equalization for Live Venue Sound Systems Damien Dooley, Final Year ECE Final Presentation, Thursday 10 th April 2008
2
Presentation Outline Project Background Major milestones and how they were achieved Conclusion
3
Project Background Difficult to get a good sound in enclosed venues Sound Engineer required to balance room acoustics Room acoustic phenomena cause impurities in sound Project aim to develop a system to counteract the negative effects of room acoustics on audio
4
The System Performer needs to convey audio clearly. Room serves to add a degree of nonlinearity to the audio. Microphone picks up room audio and feeds to DSP board together with clean sound directly from the source. DSP board attenuates irregularities to produce a cleaner sound.
5
Room Acoustic Modelling Science of determining how sound propagates though a room from source to observer. Typical impulse response of a room results in a series of exponentially decaying reflections, known as reverberation. Impulse response graph showing room reflections. Time taken to decay can be calculated using Sabine’s equation for reverberation time.
6
MATLAB MATLAB was used to create an accurate model of the room environment. Firstly a delay was created and then a natural sounding reverb as follows. num=[0.8,zeros(1,3000),1]; den=[1,zeros(1,3000),0.8]; d1=filter(num,den,d);
7
Utterance of the word “Reverberation Algorithm” prior to processing. Exponential falloff in Peaks indicate reverberation being applied to data.
8
Next Step Model now created to show reverberation Next step is to create a filter to eliminate such reverberation An adaptive filter is required for such a task
9
Arrangement of the adaptive filter within the system
10
Aside: Simulink Simulink is an extension to MATLAB and uses a more graphical approach to system design
11
Adaptive Filter in MATLAB Initial code contained the LMS adaptive filter as an object ha = adaptfilt.lms(taps,mu); [y,e] = filter(ha, b, a); Step size mu determines rate of convergence of adaptive filter towards ideal coefficients Adaptive filter is a type of filter that adjusts it’s coefficients according to the input
12
Initial Result Original Signal Reverberated Signal Filtered Signal
13
Better Understanding of the LMS r(i) = h(i)*x(i); e(i) = y(i) - r(i); h(i+1) = h(i) + (2*mu*e(i)*x(i)); Sample test data (non-audio) was applied to the following algorithm
14
Step size 2.5 x 10 -6 Step size 5 x 10 -6
15
Real Time Adaptive Filtering
16
Initial Delay Code for(msec = 0; msec < 5000; msec++) { for (sample = 0; sample < 48; sample++) { while(!DSK6416_AIC23_read(hCodec, &data)); audio[sample] = data; delayAudio[(msec+delay)%5000][sample] = audio[sample]; EchoAudio[msec][sample] = audio[sample]+0.5*(delayAudio[msec][sample]); while (!DSK6416_AIC23_write(hCodec, EchoAudio[msec][sample])); } Circular Buffer Usage
17
Real Time LMS Filter Test r[0] = 0 * 100 = 0 b[0] = 2 – 0 = 2 h[1] = 0 + (2 * 0.00001 * 100 * 2) = 0.004
18
Audio Test Output values Converging Coefficients
19
Conclusion Project aim to develop a system to counteract the negative effects of room acoustics on audio. Achieved to a substantial extent, project capable of cancelling delay.
20
Potential Improvements Project can be extended to include reverberation 2 independent signals can be used Algorithm can be tested in a real live environment
21
Thanks!
22
Questions…
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.