Conlang phonologies in SynthV
Note: needs updating for SynthV Studio
Recently, I've been playing around with SynthV because it runs on (GNU/)Linux natively. Being the conlang junkie, I've decided to try making it sing in conlangs.
Note about voicebanks
Currently [as of 23X2019], there are eight SynthV voicebanks available:
- Eleanor Forte for English
- Yamine Renri and Genbu for Japanese
- AiKO, Chiyu, Shian, Cangqiong and Haiyi for Mandarin Chinese
All languages support phonetic input:
Start the lyrics with a full stop and separate each sound with a space. Phonetic input will be indicated in
m aa n ow s p ey s.
Generally, how well a voicebank can emulate the phonology of a given language depends on the sounds available. Phonotactics is usually not an issue – even Japanese voicebanks such as Renri can handle consonant clusters.
|The sound can be emulated well enough that differences are hardly noticeable.|
|The sound has a close equivalent in voicebanks of the given language.|
|Emulation is difficult. Failure to distinguish the phoneme in question from another in the same language warrants this colour or worse.|
- As of the production release, final consonants of a syllable will
not be pronounced in Japanese voicebanks if the syllable is spread across multiple notes with a legato.
Workaround: use two separate syllables on the respective notes; e.g. instead of
.m a r z; -use
.m a; .a r z.
- In fact, Japanese voicebanks seem to have more trouble with consonant clusters in the full version than in the technical preview. If you encounter this problem, adding auxiliary syllables and adjusting voicing as needed might help.
- It is possible to combine multiple voicebanks from different languages to "iron out" each other's weaknesses.
- How to distinguish various rhotics and [l] from each other better in Japanese voicebanks?
- How to reproduce [ɮ x ɣ ø y]?
- How to reproduce ejectives and implosives?
- Possible to get better vowels in English voicebanks?
- How well would Mandarin voicebanks fare?