Fricative sounds f v s z x h like other speech sounds involve a sound source or sources and a filter function We ll start with a demonstration of how to make a voiceless fricative from a source and a filter Source turbulent noise Filter a resonator Here s some pure white noise This is the SOURCE of acoustic energy for our synthetic fricative noise In the vocal tract noise like this is made by turbulent airflow more later And here s a spectrogram of that white noise equal amplitude at all frequencies Now we apply a band pass filter center frequency is 7kHz and bandwidth is 1 kHz It is a noise formant to anticipate this 7 kHz formant is one of the resonances of the front cavity of the vocal tract That long filtered noise may not have sounded much like a fricative noise to you Check out how it sounds when we take a small chunk of filtered noise and splice it before a vowel Let s do that again with a different filter function We start again with the sample of white noise and this time filter a band centered on 2 kHz Again splice a small snippet of this noise on a vowel and we have a fricative sound Can you identify it s place of articulation So far we have an acoustic description of fricatives Source noise Filter band pass at different frequencies Let s now relate these two acoustic properties to physical properties in fricative production Where does the noise for fricatives come from turbulent motions of air particles two ways to get turbulence in the vocal tract 1 force air through a narrow opening 2 blow air on an obstacle like the teeth Regarding turbulence generated by forcing air through a narrow opening Turbulence is a function of two things 1 Volume velocity of air passing through 2 Size of the channel A narrow channel will generate turbulence more easily at lower volume velocity than will a wide channel compare different straw diameters to test this out Regarding obstacle versus channel fricatives The obstacle fricatives involving a jet of air directed onto the upper s or lower teeth are louder than the channel fricatives Obstacle fricatives are also called sibilant fricatives Two more observations about the source in fricatives 1 It isn t white noise Amplitude drops as frequency goes up Two more observations about the source in fricatives 2 The loudness of the turbulent noise depends very much on the area of the constriction The solid line shows the amplitude of fricative noise as the constriction goes from 0 to 0 5 cm2 Now for some observations about the filter function for fricatives 1 The filter for most fricatives is the front cavity of the vocal tract Narrow channel prevents acoustic coupling with back cavity fn 2n 1 c 4L 2 A simple uniform tube model does pretty well at predicting the locations of peaks in fricative spectra fricative spectra for five different places of articulation 0 lips 5 9 velar 2 A simple uniform tube model does pretty well at predicting the locations of peaks in fricative spectra e g a fricative with a front cavity length of 2 2 cm should have a formant frequency resonance peak at 3977 Hz f c 4 2 2 3977 fricative spectra for five different places of articulation 0 lips 5 9 velar 3 Labial fricatives don t have much of a vocal tract filter turbulence produced at the lips notice the 0 curve in the preceding graph 4 s and very similar place of articulation very different formant frequencies Lip rounding and sublingual cavity give a much longer front cavity