Jastiar (Indonesia), A.H. Abdullah, and D. Mohamad (Malaysia)
Speaker identification, discrete Fourier transform, template matching.
System protection that uses passwords or codes can be guessed or replicated, thus making the system vulnerable. In a speaker identification or verification, the prime interest is not in recognizing the words but determining who is speaking the words. We try to identify the speaker in a noisy room, by implementing the discrete Fourier transform (DFT) for finding frequencies buried in noise. Ten frequencies, which have the highest power spectral density (PSD), are classified as speech pattern. Template matching and Euclidean distance are applied to recognize these patterns. The system that we have developed could identify 50 utterances with accuracy more than 90 percent.
Important Links:
Go Back
