self-calibrating audio signal equalization

Self-Calibrating Self-Calibrating Audio Signal Audio Signal Equalization Equalization

Greg BurnsGreg BurnsWade LindseyWade Lindsey

Kevin McLanahanKevin McLanahanJack SametJack Samet

2Group 15 - Wade Lindsey

Project ScopeProject Scope

In any closed room, standing waves exist In any closed room, standing waves exist that change the way audio signals arrive that change the way audio signals arrive at the ear.at the ear.

Variations in amplifier design, speaker Variations in amplifier design, speaker efficiency, and room geometry affect the efficiency, and room geometry affect the frequency response, degrading it from frequency response, degrading it from flat-band operation.flat-band operation.

The goal of this project is to automatically The goal of this project is to automatically calibrate an audio signal to compensate calibrate an audio signal to compensate for these effects.for these effects.

3Group 15 - Wade Lindsey

Audio FundamentalsAudio Fundamentals

Pink Noise is a randomly generated Pink Noise is a randomly generated signal that exhibits a constant voltage signal that exhibits a constant voltage per octave.per octave.

A spectrum analyzer can be used to A spectrum analyzer can be used to obtain the actual frequency response of obtain the actual frequency response of an audio signal when placed in a test an audio signal when placed in a test position in a room.position in a room.

A graphic equalizer can then be used to A graphic equalizer can then be used to adjust the amplifier input to compensate adjust the amplifier input to compensate for any deviations off flat-band responsefor any deviations off flat-band response

4Group 15 - Wade Lindsey

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

5Group 15 - Wade Lindsey

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

6Group 15 - Wade Lindsey

Equalizer SpecificationsEqualizer Specifications

10 Bands (32, 64, 128, 256, 512, 10 Bands (32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384 Hz)1024, 2048, 4096, 8192, 16384 Hz)

Filters, input, and output Filters, input, and output constructed using LM351 op-ampsconstructed using LM351 op-amps

Discrete components and 10kΩ Discrete components and 10kΩ DS1803 digital potentiometersDS1803 digital potentiometers

7Group 15 - Wade Lindsey

10-Band Equalizer 10-Band Equalizer CircuitCircuit

8Group 15 - Wade Lindsey

Built EqualizerBuilt Equalizer

9Group 15 - Wade Lindsey

Equalizer ResponseEqualizer Response

Frequency response of equalizer with varying resistances tested with HP VEE.

10Group 15 - Kevin McLanahan

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

11Group 15 - Kevin McLanahan

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

12Group 15 - Kevin McLanahan

Pink Noise SpecificationsPink Noise Specifications

Creates pseudorandom digital noise Creates pseudorandom digital noise for white noise in first stage at for white noise in first stage at 3dB/dec3dB/dec

Second stage pink noise filter at -Second stage pink noise filter at -3dB/dec3dB/dec

Frequency response 20 Hz – 20 kHzFrequency response 20 Hz – 20 kHz 33-bit resolution in shift register for 33-bit resolution in shift register for

pseudorandom number generationpseudorandom number generation Line level output at 150mV rmsLine level output at 150mV rms

13Group 15 - Kevin McLanahan

Pink Noise GenerationPink Noise Generation Equal voltage per octave across audio band.

FFT of Pink Noise viewed on oscilloscope.

14Group 15 - Kevin McLanahan

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

15Group 15 - Kevin McLanahan

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

16Group 15 - Kevin McLanahan

Spectrum AnalyzerSpectrum Analyzer

Samples input signal from microphoneSamples input signal from microphone Performs an FFT (Fast Fourier Performs an FFT (Fast Fourier

Transform) algorithm to extract Transform) algorithm to extract frequency componentsfrequency components

Compares relative frequency levels to Compares relative frequency levels to optimal flat-band responseoptimal flat-band response

Samples microphone input at 19.2 μsSamples microphone input at 19.2 μs Sample length of 256 data points at 8-Sample length of 256 data points at 8-

bit resolutionbit resolution

17Group 15 - Kevin McLanahan

FFT ExplainedFFT Explained

Implementing Cooley/Tukey FFT algorithm.Implementing Cooley/Tukey FFT algorithm. Has Big O of N log NHas Big O of N log N Takes Fourier matrix of power 2 (2Takes Fourier matrix of power 2 (288 in our in our

case)case) Breaks into 2 log N matrices and performs Breaks into 2 log N matrices and performs

multiplications on roots of unity (emultiplications on roots of unity (e2πihk/N2πihk/N)) Ultimate result returns a vector with Ultimate result returns a vector with

frequency, phase, and magnitude frequency, phase, and magnitude information.information.

18Group 15 - Greg Burns

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

19Group 15 - Greg Burns

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

20Group 15 - Greg Burns

Microcontroller SoftwareMicrocontroller Software

Initialization Mode

Calibration Mode

Operation Mode

Sets all initial variables and default settings.

Outputs pink noise signal through speakers, receives spectral data from analyzer,

and adjusts equalizer to compensate.

Resets MUX to audio source and selects current room location to compensate.

21Group 15 - Greg Burns

Initialization ModeInitialization Mode

Sets system timers and interruptsSets system timers and interrupts Configures input and output pinsConfigures input and output pins Defaults variables to initial Defaults variables to initial

conditionsconditions Initializes IInitializes I22C transfersC transfers Sets digital pots to a predetermined Sets digital pots to a predetermined

ideal flat-band responseideal flat-band response

22Group 15 - Greg Burns

Calibration ModeCalibration Mode

Gathers data from spectrum analyzer Gathers data from spectrum analyzer outputoutput

Compares current frequency response Compares current frequency response peaks to ideal responsepeaks to ideal response

Adjusts digital potentiometers based upon Adjusts digital potentiometers based upon previous comparisonprevious comparison

Repeats until current frequency response Repeats until current frequency response and ideal response fall within 5% toleranceand ideal response fall within 5% tolerance

Calibrates for every room positionCalibrates for every room position

23Group 15 - Greg Burns

Operation ModeOperation Mode

Switches audio source from pink Switches audio source from pink noise to preampnoise to preamp

Sets digital pots to specific values Sets digital pots to specific values corresponding to room locationcorresponding to room location

Monitors Sensor Arrays for room Monitors Sensor Arrays for room location variationslocation variations

24Group 15 - Jack Samet

Block DiagramBlock Diagram

Spectrum Analyzer

PICMicrocontroller

Pink Noise Generator

Audio Preamp

MUX

Equalizer

Sensor Array

25Group 15 - Jack Samet

Future ImprovementsFuture Improvements

Increased number of frequency Increased number of frequency bands on EQbands on EQ

Use of DSP processor for improved Use of DSP processor for improved FFT performanceFFT performance

Use of audio-grade tolerance Use of audio-grade tolerance componentscomponents

Allow for wide variety of sensor Allow for wide variety of sensor array configurationsarray configurations

26Group 15 - Jack Samet

ConclusionsConclusions

Economically feasible and marketableEconomically feasible and marketable Modular design allows for easy Modular design allows for easy

implementation, innovation, and implementation, innovation, and reproductionreproduction

Compatible with most modern stereo Compatible with most modern stereo systemssystems

Overall a universally usable product Overall a universally usable product from personal to commercial from personal to commercial applicationsapplications

27Group 15

Questions?Questions?

28Group 15

FinFin