Voice Recognition Hardware Development Read My Lips John Porter, Lavanya Mynam, Gerald Mortensen

Voice Recognition Hardware Development Read My Lips

John Porter, Lavanya Mynam, Gerald Mortensen

Problem Statement

• Next generation keyboard architecture for voice recognition:- Identify microphone candidates- Direction Finding- Basic signal Conditioning- Optional wireless input- Performance testing protocol- Integrated keyboard packaging

Project Learning: Microphones

Condenser Microphones• Electret Microphones

Omni-directional microphones Unidirectional microphones

Specifications• Impedance (about 200 Ω or lower) • Frequency response ( 50 Hz to 16 kHz)• Dynamic range (softest to loudest sound)• Sensitivity (47 dB)• Cost• Size

Project Learning: Detecting Direction

• Theory– Subtracting signals to create a highly

directional microphone.

Project Learning: Placement Idea

Project Learning: Signal Conditioning

• Single unidirectional microphone

Uni

Omni

D

Microphone Direction Testing

Project learning: Issues

• Technical questions– Precision– Microphones locations– Toggle between text and software

Project learning: Direction Finding

• Multiple unidirectional microphones

Omni

Uni

Uni

Uni

Comparators “C”Differential Op Amp “D”

D

D

D

C C

Project learning: Wireless

• Electrosonics

Project learning: Via Voice

• Platforms ( Windows Me/98/2000/XP )

• Capabilities– Converts speech to text– User adaptation– Understandability (accents)– Multiple users

Testing Ideas

• Methods– Lazy Susan– Waveform Comparisons– Speech to text translation

• Metrics– Noise level– Noise source– Microphone orientation– Accuracy

Project Management

• Schedule– Mid project review: 4/12/2007– Initial prototype: 4/25/2007– Final Design Review: 5/3/2007

Budget

• Budget ( $4000 total )– $50 Via Voice– $50 Poster– $300 USB audio device– $400 Two PCB Designs with setups– $200 Testing hardware– $750 Transportation– $500 Miscellaneous parts– $250 Miscellaneous tools– $500 Reserve– $1000 ECE Program Contribution