SYLLABLESOUNDsings syllableslight / dark

SYLLABLE

SYLLABLE sings syllables: each key plays a consonant into a vowel, and the vowel holds for as long as the key does. It is a formant synthesizer in the style of Klatt's, built for the consonants that VOICE does not have.

sings syllables

Half

SOUND

Reach it

a custom build, FELUCCA_SYLLABLE=1 (else on MOUTH)

ALGORITHM

CONS

Engine

SYLLABLE

Factory sounds

11

Playing it

Every note sings CONS into VOWEL on the key's pitch, up to four at once. Hold the key and the vowel sustains; let go and the syllable ends, closed by END if you have set one (ba with END on t sings "bat").

ALGORITHM picks the consonant. MAIN's four knobs build the syllable:

• CONS picks the consonant: stops such as b, d and k, nasals such as m and n, hissing s, z and sh, and more. The dash means none, a bare vowel.
• VOWEL picks one of Hillenbrand's twelve vowels, a diphthong (a vowel that glides into another), or iau, a cat's "mee-ow" when it follows m.
• SHIFT moves every formant (the resonances of the mouth) up or down: up for a smaller mouth, down for a larger one.
• BRTH adds breath noise under the whole syllable.

The screen spells the syllable over a vowel chart: the twelve vowels sit where their first two formants put them, the sung vowel is a dot, and a diphthong is an arrow from its first vowel to its second.

Load HELLO and play four notes in a row: its PICK is set to the HELLO phrase, so it sings one syllable per note.

SYLLABLE is on the sidelines: SQRL-1's build leaves it out. A custom build with FELUCCA_SYLLABLE=1 brings it back, picked on the PAIR page (hold HOME, tap GLO, KNOB 1). Without it, a SYLLABLE sound in a project or a saved sound plays on MOUTH.

MAIN

Pages

SYLL

LEN shortens or stretches the consonant before the vowel arrives.

Knob	Label	Range	Starts at
1	CONS	19 values - b d g p t k m n s z sh f v h l r w y	b
2	VOWEL	18 values i I e E ae a V O o U u er ai au ei oi ou iau	a
3	LEN	0% .. 100%	50%
4	SHIFT	-12st .. +12st	0st

MOUTH

PICK decides what the next note sings. FIX sings the syllable as set. C>> steps to the next consonant with each note, V>> to the next vowel, BOTH to both, and RAND picks at random. The named settings are phrases, sung a syllable at a time: HELLO sings "hel-lo maze-mind", DATABN "da-ta-burn", FM ONE "ef-em-one" and ROBOT "ro-bot". A take sings its phrase from the top each pass: on a part that holds a take or is recording one, the phrases, C>> and V>> start over at step 1, so the loop sings what you sang.

FALL lets the pitch drop over each syllable, as speech does at the end of a phrase. END is the consonant that closes the syllable when the key is let go. It plays during the release, so give the sound a release long enough to hear it (ENV, REL).

Knob	Label	Range	Starts at
1	BRTH	0% .. 100%	0%
2	PICK	FIX C>> V>> BOTH RAND HELLO DATABN FM ONE ROBOT	C>>
3	FALL	0st .. +12st	+2st
4	END	19 values - b d g p t k m n s z sh f v h l r w y	-

Where it comes from

The synthesizer follows Klatt (1980): a cascade of formant filters for the vowels and parallel ones for the consonants. The vowels' formants come from Hillenbrand et al. (1995). The C code, firmware/src/syllable.c and firmware/src/eng_syllable.c (MIT), was checked bit for bit against a reference model that is not published.