Section Handout 10/24/2008

Transcription

1 Section Handout 10/24/2008 Vowels Acoustic-Phonetics: At any given time in the production of a vowel, the spectrum of the sound radiated from the lips can be found from the product of the excitation (source) spectrum generated by the larynx and the frequency response (filter) of the vocal tract configuration. Typically the spectrum of a vowel is made up of a large number of harmonics, with those harmonics occurring at frequencies close to the resonant frequencies of the tract (formants) having the greatest amplitude. The location of formant frequencies can be reasonably well predicted from the vocal tract shape by means of a two-tube approximation. This demonstrates the observed correlation between F1 and size of cavity behind the tongue hump, and the correlation between F2 and the size of the cavity in front of the tongue hump. The large changes in F1 and F2 across vowel articulations explain why these two features of vowels have the greatest perceptual effect. [Perception: The hearing mechanism delivers to the brain of the listener a neural pattern related to the spectral envelope of the vowel spectrum. This spectral envelope varies according to both the pitch and the timbre of the sound. Since the absolute values of the formant frequencies depend on the size of the speaker s vocal tract, the listener must first compensate for vocal tract size before attempting to identify the vowel. This process is called normalisation, and can be thought of as a transformation of the actual vowel quality to the equivalent vowel quality of a standard speaker. The normalised quality can then be mapped to a position on a perceptual map (a lot like a vowel quadrilateral) and its phonetic quality retrieved with respect to known qualities.]

2

3 Spectrograms and Linear Predictive Coding spectra Average stylized formant frequencies (F1-F3) of the monophthongal vowels of American English produced by 50 male speakers

4

5

6 Approximants are non-vowel oral sonorants. In English, this amounts to /l, r, w, j/. They are characterized by formant structure (like vowels), but constrictions of about the degree of high vowels or slightly closer. (Generally there's no friction associated with them, but the underlying approximants can have fricative allophones, just as fricative phonemes can occasionally have frictionless (i.e. approximant) allophones.) Characteristics: In general, approximants are acoustically quite similar to those vowels produced at a comparable location in the vocal tract. They are distinct by being shorter and weaker than vowels and by having longer formant transition durations than vowels. - Palatal Approximant low F1, high F2 and F3, quite pronounced transitions both in frequency range and duration - Labial-velar Approximant low F1, F2, which are close together, F3 steady around 2300 Hz - Lateral Approximant low intensity, center frequencies F1 250 Hz, F Hz, F Hz, abrupt change in intensity at the beginning of a vowel - Retroflex Approximant very low F3, that comes close to F2

7

8

9 Nasals Nasal consonants involve a lowering of the soft-palate (velum) which links in the nasal cavities as additional acoustic resonators. The manner cues for nasals include the presence of a low-frequency resonance due to the nasal cavity, and the rapid fall and rise in energy as the nasal is made and released. The place cues to nasals mostly arise from the second and third formant transitions, as for plosives. Nasals have identifiable formants that are lesser in amplitude than in the vowel, and the regions between them are blank. In addition, the spectral shape of the nasal itself varies slightly with the place of the obstruction in the vocal tract. This seems to be due to the size of the cavity trapped behind the obstruction which modifies the filter characteristic of the branched tube. The nasalisation of vowels is cued by the presence of a low-frequency resonance and an increase in formant damping. Characteristics - low F1 - F are weak, i.e. low in amplitude - increased bandwith - anti-formants Interaction of oral and nasal cavity leads to anti-formants, i.e. those resonant frequencies in the nasal that are close to resonances of the oral cavity are absorbed or canceled and subtracted from the overall spectrum because part of the acoustic energy of the pulsating airflow from the larynx causes resonance in the oral cavity - spectral properties of the transition into and out of surrounding vowels

10 Spectrograms of "dinner", "dimmer", "dinger"

11

12 Fricatives, by definition, involve an occlusion or obstruction in the vocal tract great enough to produce noise (frication). Frication noise is generated in two ways, either by blowing air against an object (obstacle frication) or moving air through a narrow channel into a relatively more open space (channel frication). In both cases, turbulence is created, but in the second case, it's turbulence caused by sudden 'freedom' to move sideways. Sibilant fricatives involve a jet of air directed against the teeth. While there is some (channel) turbulence, the greater proportion of actual noise is created by bouncing the jet of air against the upper teeth. The result is very high amplitude noise. Non-sibilant fricatives are more likely 'pure' channel fricatives, particularly bilabial and labiodental fricatives, where there's not a lot of stuff in front to bounce the air off of. Characterized in terms of four attributes: - spectral properties of the frictions noise - amplitude of the noise - duration of the noise - spectral properties of the transition into and out of surrounding vowels Characteristics: - Voiced fricatives two energy sources: turbulent noise from constriction, low-frequency energy from the vibrating vocal folds - [s, z, S, Z] quite pronounced clear distinct spectral shape, greater friction noise and longer duration than [f, v, T, D] - [S, Z] mid frequency around Hz - [s, z] higher spectral peak frequency than [S, Z], around [f, v, T, D] relatively flat spectrum - [f, v] higher spectral peak frequency than [T, D], around Velar, uvular, pharyngeal fricatives energy is concentrated at lower frequencies, in the area of the first two formants - Velar fricatives concentration of energy in the region corresponding to F2 of the adjacent vowel, little energy in the higher frequencies - Uvular, pharyngeal fricatives main difference between uvular and pharyngeal fricatives seems to be in terms of F1 frequency - pharyngeal fricatives have a higher F1 than uvular fricatives - Formant transition F2 at the onset of the vowel tends to increase as place of articulation of the fricative moves further back in the oral cavity

13 Top row, left to right: f, theta, s, esh. Bottom row, left to right: v, eth, z, yogh

14

15

16 Plosives Acoustic Cues: when listeners interpret a speech signal, they look for aspects of the spectrotemporal pattern that indicate which particular phonetic and phonological components were produced by the speaker. In terms of vowels, we have seen that the identity of the vowel is indicated (or cued) by the formant frequencies. For fricatives, we have seen that the frequency of the main spectral peak and the bandwidth of that peak increase as the place of constriction moves further forward in the mouth. For diphthongs and approximants, we have seen that their identity is indicated by both the frequencies and the shapes of formant movements. In general the manner of obstruents is indicated by rapid formant transitions in the vowel regions leading up to and following the articulation of the obstruent, particularly the first formant transition. The place of obstruents is also indicated in these formant transitions, particularly the shape of the second and third formant transitions. Since transitions change in shape a great deal according to the identity of the vowel, we introduced the notion of locus frequency to independently characterise the transition shape for a given consonant. All of these indicators: formant frequencies, spectral peaks, transitions, locus frequencies are called acoustic cues to phonetic identity. Plosives: the articulation of a plosive requires a closing articulation phase, an obstruction phase (stop gap), a release phase, an optional aspiration phase, and an opening articulation phase. These phases have characteristic acoustic cues associated with them. The manner cues for plosives include the presence of the silent region in the stop gap, the rapid formant transitions and particularly a low locus frequency for F1, sudden energy change, release burst and aspiration. The place cues for plosives include the centre frequency (i.e. main spectral peak) of the turbulence occurring at the release (the burst), and the locus frequency for the second and third formant transitions. The burst centre frequency cue turns out to be processed relative to the frequency of the vowel F2. Place Burst Centre Frequency F2 Locus Frequency F3 Locus Frequency Bilabial Lower than vowel F2 Low Low Alveolar Higher than vowel F2 Mid High Velar Close to vowel F2 High Mid The voicing cues for plosives include the voice onset time, the presence of aspiration, the presence of an audible F1 transition, the intensity of the burst and the duration of the preceding vowel. There are notable differences in cues to voicing across languages: some do not use aspiration, others have a three-way contrast.

17

18

19 Spectrograms of "bab" "dad" and "gag"

20

21 Affricates