| Summary: | Microphone array and beamforming techniques can be used for eavesdropping or listen to people talking at a certain distance away from the microphone. Using multiple microphones, it is possible to get spatial information from the phase difference in sound waves and perform different type of signal processing when the waves reach the microphones. [1] In this project, speech recognition system based on ad hoc microphone array using blind source separation technique will be observed. A mixture of voices will be recorded by multiple speakers. Even though the signals received by different recording devices are not synchronized, the blind source separation technique will compensate the differences in sampling frequency mismatch and time offset. This project also observes the maximum likelihood estimation that that will be obtained by a golden section search. The main programming software for this project used will be MATLAB, which the author is expected to implement individual recordings while taking that the sources to be motionless and have stationary amplitudes. Areas for future research will also be proposed in this report.
|
|---|