We demonstrate the use of this framework in recognizing speech from mixtures of known speakers.
The observed signal is a mixture of the two speakers z[n]=a[n]+b[n], (2).
Figure 9 Modulation spectrogram and segments obtained for a mixture of (a) male speaker, (b) white noise, and (c) trill telephone.
Mixture Mix. 4 Mix. 5 Number of speakers 2 3 Speaker Codes cp,ed fm,ga,eDistancece between speaker and array center m m Angular position of the speaker (as shown in Figure 4) Figure 4 Experimental Setup.
Mixture Mix. 1 Mix. 2 Mix. 3 Number of speakers 2 3 2 Speaker Codes ar,ed pg,ed,cp fm,pg Distance between speaker and array center m m m Angular position of the speaker (as shown in Figure 4) Table 2 Mixture description.
While we do not have a definitive answer to why there is the discrepancy between Mandarin Chinese and other languages with regard to name agreement, our speculation is that Mandarin Chinese, especially that used in Mainland China, is often a mixture from speakers of many different dialects, as well as many ethnic groups.
All of the mixtures were made with the same number of microphones as the number of speakers in the mixture (M = N), i.e., in each mixture a determined problem is considered so the classical ICA algorithms for source separation can be applied.
In [19], we introduced a post-filtering method which is implemented after an optimum beamformer to extract the desired speech source from a mixture of signals in multiple speakers environments.
were used, where N is the number of speakers in the considered mixture.
In a UBM-GMM system, each speaker's distribution of feature vectors is modeled as a mixture of Gaussians, forming a Gaussian mixture speaker model (GMSM).
