|
Create a new KlattGrid from the specifications for a vowel.
The following script creates a vowel sound with many formants which will sound like the vowel /a/. The formant frequencies will be: F1 = 800 Hz, F2 = 1200 Hz, F3 = 2300 Hz, F4 = 2800 Hz. The frequencies of the higher formants will be at intervals of 1000 Hz, starting from F4. Therefore, F5 = 3800 Hz, F6 = 4800 Hz, F7 = 5800 Hz, and so on. The bandwidths will be B1 = 80 Hz, B2 = 80 Hz, B3 = 100 Hz. The bandwidths of the fourth and higher formants will be 0.05 times their frequency. Therefore, B4 = 140 Hz, B5 = 190 Hz, B6 = 240 Hz, B7 = 290 Hz, and so on.
Create KlattGrid from vowel: "a", 0.3, 125, 800, 80, 1200, 80, 2300, 100, 2800, 0.05, 1000
To Sound
The following script will create a two formant sound which also sounds like an /a/.The formant frequencies will be 800 Hz and 1200 Hz.
Create KlattGrid from vowel: "a", 0.3, 125, 800, 80, 1200, 80, 0, 100, 0, 0.1, 0
To Sound
The following script will create a formant sound which also sounds like an /a/. The formant frequencies will be 800 Hz, 1200 Hz, 2200 Hz, 3200 Hz, 4200, and so on, with 1000 Hz separation).
Create KlattGrid from vowel: "a", 0.3, 125, 800, 80, 1200, 80, 0, 100, 0, 0.05, 1000
To Sound
Because all the frequency points of the corresponding tiers in this KlattGrid are defined at the start time of the grid, i.e. at time 0.0 seconds, it is easy to change the characteristics of the vowel sound by adding new points. For example, given one of the /a/ sounds above which were all synthesized with constant pitch we can have a falling pitch with:
Create KlattGrid from vowel: "a", 0.3, 125, 800, 80, 1200, 80, 2300, 100, 2800, 0.05, 1000
Add pitch point: 0.3, 100.0
To Sound
An /au/ diphthong is also easily made by a simple extension with two oral formant frequency points:
Create KlattGrid from vowel: "au", 0.3, 125, 800, 80, 1200, 80, 2300, 100, 2800, 0.05, 1000
Add pitch point: 0.3, 100.0
Add oral formant frequency point: 1, 0.3, 300
Add oral formant frequency point: 2, 0.3, 600
To Sound
Formant frequencies and bandwidths for 16 Swedish vowels are presented by Hawks & Miller (1995) in their table 1. They further give equations for bandwidths as a function of formant frequency. Fleisher et al. (2015) present in their table 1 values for the first 5 formant frequencies and bandwidths of the German vowels /a/, /i/ and /ʊ/ for speech mode and for singing mode. In general their bandwidths are lower than as predicted by the Hawks and Miller equations.
© djmw 20230801