Search results

Ramadan Kareem!

I once not really fond of homebrew AI voicebanks, mainly because of the usability, but now people are making it more usable than ever and I was like 🤯 this is what I really want. But yeah, making them is still not as effective as making concatenative voicebanks.

Dreamtonics singlehandedly killed Append™ voicebanks.

The person I discussed with tried it with Stardust and Tsuina-chan, and sounds like they're all not sensitive with the parameter. Well, seems like the parameter is somewhat picky to the databases.

Too bad that I can't try SynthV Studio Pro and full AI databases. So there's this discussion on a Discord server that the Tone Shift parameter is actually doing nothing, in the sense of the falsetto part can still be heard in the falsetto notes even if you set the parameter low. I was like "eh...

I often heard the allegations on Twitter that Japanese companies are somewhat racist by not cooperating with Chinese companies/people. I don't buy that. Just look at Hatsune Miku V4C.

Back in the days everyone and their grand mom wanted to be Hatsune Miku, and now guess like everyone and their grand mom are wanting to be Kafu.

Said concept would theoretically works. But, if you think about it, wouldn't it makes the Tension parameter—one of the main selling point of the engine—essentially useless and redundant? Though we are still waiting for the Glottal effects redesign Sure it would be easier if we want specific...

ヨシ! 👈😺

Sakakibara tweeted that it's new recordings. Very likely what they meant of "special technique" is some sort of "fluent-izer" algorithm (similar technique with use case scenario of making Chinese-accented Japanese speech more fluent), which is should also be cool. What bugs me the most is, why...

Surprised and disappointed that there are no research whatsoever, let alone a record, about fastest tempo human can sing.

One of the essence of AI is to mimic, or precisely, reflect human beings. If an AI is racist, it's the human that is racist. If an AI can't sing in extreme tempo, it's the human that can sing in extreme tempo.

After listening to IA AI demos, I really want to try CeVIO AI whenever I can afford it, for the sake of science.

I never want to shop online, never will be. I would rather shop face to face with the goods directly present in front of me. I/you thought that I'm not an e-commerce platforms' target audience, but that's wrong. I still get bombarded by those platforms' ads everywhere I go. Also the same with...

People: We lose our accent when we sing. Virtually any person involved in voicesynth community/industry: No, we didn't.

Tried SynthV Studio Basic on my aunt's laptop, and, man, it's so much fun to use! It's ridiculously snappy even though her laptop is underpowered to my standard. Got zero problem using it. Thank you Dreamtonics, very cool.