1Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Team Acoustic Beamformer

Preliminary Design Review

10/18/2013

2Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Team Acoustic Beamformer

Nick DriscollEE

Rebecca McFarlandCSE

John ShattuckEE

Jimmy DanisEE

Name

3Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Presentation Overview• Problem Statement• Social Impact• Our Project: The Acoustic Beamformer

• System Requirements• Block Diagram• Description of Components

• Alternatives• Current Development Status• MDR Deliverables• Schedule

4Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Problem Statement• Difficult to scan and localize a single person’s voice

among outside conversation and background noise in real time

5Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Problem Areas• Primary: Conference

Rooms/Small Lecture Halls

• Video conferences across business sites

• Questions asked in lecture

• Secondary: Surveillance

• Detecting noises in small rooms

• Would be nice for surveillance cameras to quickly point to an intruding noise source

6Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Social Relevance• Hard for those with hearing impairments to filter

out background noise

• Cost considerations

• High-end hearing aids can cost up to $8000

• Up to 75% of hearing-impaired individuals do not have hearing aids

7Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Solution: The Acoustic Beamformer

Input: 8 microphone outputs

Output: Project audio signal and visual representation of sound wave

Utilize Beamforming Signal Processing Techniques

8Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

System Requirements• Operate within the human voice frequency

spectrum

• Ideally 300 Hz- 3kHz

• Localize a sound source within 5 meters of the microphone array (15 feet)

• Effective in an 100 degree span15 ft.

40o(Not to Scale)

9Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Final System FunctionalityInitial Goal:

• Manually input desired target angle into the PC

Optimal Goal:

• Scan, find and fixate on individual sound sources within our 100o span

• Alternative: Identify an electronic “bug” emitting a high frequency sound (15kHz) to fixate on

10Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Block Diagram

11Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Microphones• Analog Device MEMS Microphone

• Omnidirectional

• Analog output

• Frequency range: 100 Hz – 15 kHz

• Sensitivity -42 dB +/- 3 db @ 94 dB SPL

• S/N Ratio 62 dB

12Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

A-D Converter• Need 8 channel inputs

• Need USB out because modern computers do not support serial interfaces

• Options we are investigating:

MC USB-DIO24/37 DATAQ DI-149 USB

13Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Computer Software• MATLAB for simulation and initial analysis

• Custom software for real-time processing

• DSP processing library

• Visualization

• Audio output

14Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Alternatives: Nontechnical• Direction of interest manually fixated

• Pass around a microphone

• Physically turn a microphone or camera to target point

• Cheap and accessible

Inefficient, more time consuming than electronic methods

15Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Alternatives: Technical• ClearOne non-directional 24 microphone array

for conference rooms

• Price: $3,000

• Polycom HDX Ceiling Microphone Array

• Price: $1,200

Systems are expensive, far exceeding an SDP budget

16Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Current Development Status• Researched similar previous SDP projects

• 3 main issues keeping others from succeeding:

• Problems integrating A/D Converters

• Choosing substandard microphones

• Assuming plane waves (sources likely too close)

• Compiled several MATLAB simulations

• Refining a basic algorithm for use

• Purchased a few microphones for initial testing before deciding on final hardware

17Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

MATLAB Plots

18Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

MATLAB – Source Angle Sweep

19Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Proposed MDR Deliverables

• Single microphone to A-D

• Input into MATLAB

• Be able to analyze one channel in MATLAB

• Parallel development of real-time software framework

20Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

ScheduleTask Area Responsible Due Date

Test mics - determine if they meet specs Hardware Nick 18-Oct

Choose A-D Hardware Nick 1-Nov

Initial MATLAB simulations MATLAB Jimmy 6-Oct

MATLAB Parameter Sweep MATLAB Jimmy 6-Oct

MATLAB Parameter Optimization MATLAB Jimmy 10-Oct

MATLAB Source Scan MATLAB Jimmy 18-Oct

MATLAB Source Track MATLAB Jimmy 18-Oct

Find DSP library Software Rebecca 18-Oct

Design software structure Software Rebecca 18-Oct

Test A-D into computer Hardware Nick 8-Nov

Process signal in software Software Rebecca 1-Nov

Process multiple inputs in software Software Rebecca 15-Nov

Get signal->computer & see in MATLAB MATLAB General MDR

MDR Review Week Misc 18-Nov

MDR Draft Report Documentation Jimmy 27-Nov

MDR Final Report Documentation John 13-Dec

Algorithm Design Misc John Ongoing

21Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Backup Information

22Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Mic Separation Sweep

23Department of Electrical and Computer EngineeringDepartment of Electrical and Computer Engineering Advisor: Professor Zink

Source Frequency Sweep