We're currently having issues with our e-mail system. Anything requiring e-mail validation (2FA, forgotten passwords, etc.) requires to be changed manually at the moment. Please reach out via the Contact Us form if you require any assistance.
The problem lies with SynthV itself. I barely know anything about SynthV or signal processing, but according to Jordi Bonada's 2008 PhD thesis, the original Vocaloid engine did most of its heavy lifting in the frequency domain. This means that an inverse fast Fourier transform was used to push the data into the time domain so that it can be turned into PCM data. Perhaps SynthV also does its heavy lifting in the frequency domain.
A quick google for "inverse fast Fourier transform" and "aliasing" together does return many results for papers on the subject of aliasing and FFTs, but it's all Greek to me.
I also downloaded Bonada's original thesis and, and a search for "alias" brings up four results. It seems that he was worried about aliasing, but not too much.
Singing voice is one of the most challenging musical instruments to model and imitate. Along several decades much research has been carried out to understand the mechanisms involved in singing voice production. In addition, from the very beginning of the sound synthesis techniques, singing has...
In that Wikipedia aliasing article there is a sound sample containing a sawtooth wave played at 440 Hz (A4), 880 Hz (A5), and 1760 Hz (A6). Each pitch is played twice, first with bandlimiting (anti-aliasing), and then without bandlimiting (aliased). The 440 Hz non-bandlimited sawtooth is closer in range to Saki's higher usable notes, and its aliasing sounds suspiciously similar to the harshness I hear in Saki. (Or SynthV in general.) Of course, the overtones of a sawtooth are inherently harsher than a voice, so the Wikipedia example exaggerates the effect, but that's probably why they used a sawtooth.
Note that I'm not a math person, but the concepts discussed aren't entirely alien to me.
And while we're on the topic of aliasing, here's an awesome Dan Worrall video about oversampling and aliasing that you may find practical when mixing. (And it might clear up some of the nonsense I wrote above):
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.