Department of Electrical and Computer Engineering and School of Computer Science Carnegie Mellon University Pittsburgh, Pennsylvania 15213
Carnegie Mellon Slide 2 18-791 Digital Signal Processing I The original sound and its spectrogram Time 0 0.2 0.4 0.6 0.8 1 1.2 0 1000 2000 3000 4000 5000
Carnegie Mellon Slide 3 18-791 Digital Signal Processing I Downsampling the waveform Downsampling the waveform by factor of 2: 0 10 20 30 40 50 60 70 80 90 100 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 n 0 5 10 15 20 25 30 35 40 45 50 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 n
Carnegie Mellon Slide 4 18-791 Digital Signal Processing I Consequences of downsampling Time 0 0.1 0.2 0.3 0.4 0.5 0.6 0 1000 2000 3000 4000 5000 Original: Downsample Downsampled:
Carnegie Mellon Slide 5 18-791 Digital Signal Processing I Upsampling the waveform Upsampling by a factor of 2:
Carnegie Mellon Slide 8 18-791 Digital Signal Processing I Filter 1 in the time domain 0 20 40 60 80 100 120 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 n 0 20 40 60 80 100 120 -2 0 2 4 6 8 x 10 -3 n
Carnegie Mellon Slide 9 18-791 Digital Signal Processing I Output of Filter 1 in the frequency domain Time 0 0.2 0.4 0.6 0.8 1 1.2 0 1000 2000 3000 4000 5000 Original: Lowpass:
Carnegie Mellon Slide 10 18-791 Digital Signal Processing I Filter 2 in the time domain 0 20 40 60 80 100 120 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 n 0 20 40 60 80 100 120 -0.01 -0.005 0 0.005 0.01 n
Carnegie Mellon Slide 11 18-791 Digital Signal Processing I Output of Filter 2 in the frequency domain Time 0 0.2 0.4 0.6 0.8 1 1.2 0 1000 2000 3000 4000 5000 Original: Highpass:
Carnegie Mellon Slide 12 18-791 Digital Signal Processing I The source-filter model of speech A useful model for representing the generation of speech sounds: Pitch Pulse train source Noise source Vocal tract model Amplitude p[n]
Carnegie Mellon Slide 13 18-791 Digital Signal Processing I Separating the vocal-tract excitation from the filter