Acoustic pattern registration for a new type of human-computer interface
D.T. Pham a, Z. Wanga , Z. Ji a, M. Yanga, M. Al-Kutubia, S. Cathelineb
aCardiff University, UK
bLaboratoire Ondes et Acoustique, ESPCI, France
Outline
- Introduction
- Principle
- Experiment
- Analysis
- Demonstration
- More information
Introduction
- TAI-CHI: Tangible-Acoustic Interfaces for Computer Human Interaction
- Current CHI hardware
- Tangible (Touchable)
- Majority
- Example: Mice, Keyboards, Touch Screen
- Intangible (Non-touchable)
- Limitation of Current Tangible Devices
- Tangible Devices restrict the mobility of users
- Users must be in certain locations during interaction
- and certain devices
- Touch-screen is normally
- Small size of computer screen
- Expensive
Application examples of tai-chi
- Interactive windows in shops or public services
- Low cost PC and video game interfaces
- Large interactive screens for academic presentation
- Novel musical instruments
- Novel interactive artistic interfaces, such as interactive painting
Solutions
- Time-delay of arrival (TDOA)
- Using a number of sensors
- Computing the differences in the arrival time of acoustic signals
- Location pattern matching (LPM)
- One or more sensors
- Localised by matching the recorded signal patterns
Principle
The LPM method is following a few steps below:
- Initilisation stage
- Vibration generated by finger tapping
- Signal conditioning
- A/D conversion
- Acquisition of template signals
- Comparison
- Compare the new signal with the recorded templates
- Correlation coefficient is used to determine the level of similarity
Experiment
- Conversion of a glass panel into an acoustic sensitive surface.
Compare difference sampling rates
The same signal pattern with three different sampling rates.
Analysis 1
a. Correlation coefficients distribution when sampling rate is 1kHz.
b. Standard deviations of correlation coefficients with sampling rates from 1 KHz to 31 KHz.
Compare differente resolutions
The same signal pattern is digitilised with different resolutions: 16bit, 12bit and 8bit.
Analysis 2
a. Correlation coefficient distribution when digitisation resolution = 8 bits
b. Standard deviation of correlation coefficients with digitisation resolution levels between 8 and 16 bits.
Compare different sample lengths
Three recorded acoustic signal patterns with different signal lengths.
Analysis 3
a. Correlation coefficient distribution when sample length = 5 ms
b. Standard deviation of correlation coefficients with sample lengths from 5 ms to 60 ms.
Demonstration
Please click on the links below to see the video demonstration.
Demonstration
Please click on the links below to see the video demonstration.
