Yes, there's a Sayaka AI demo in there (though I don't know if there's more than that one off the top of my head)
Other Software Mikoto Studio Discussion Thread
- Thread starter sapplerx
- Start date
I looked and that's all I could find. I can't share it here however because the demo is supposed to be exclusive to their Discord in case that's actually not how the final AI engine sounds (things are subject to change for reasons such as performance).
As of right now, there's one Sayaka demo available in our Discord server yes! We are currently working with a number of producers to create new songs to showcase our engine properly. Once these are ready, we'll be able to post some public demos of our voices (they're turning out so good and I cannot wait to show them off 
).
For now, please refer to the Discord server if you want to hear Sayaka in action. News is first shared there, and we try to keep everyone updated as much as we can.
).For now, please refer to the Discord server if you want to hear Sayaka in action. News is first shared there, and we try to keep everyone updated as much as we can.
A public demo for Mikoto Studio will be available on April 1st! (not a prank)
This demo comes alongside their crowdfund announcement. Particularly in light of Maghni AI, it's worth noting that significant features are almost always cut before the final release of a crowdfunded product (when they're fortunate enough to release), so donate with the understanding that no promise is guaranteed. Companies, especially when it's their first commercial release, cannot accurately anticipate the potential pitfalls they will face, and as a result almost always overestimate what they can achieve with the time/money.
Mikoto Studio lists plans to produce virtual instruments, audio effect plugins, music hardware, live performance features, and six languages: it's unlikely this will all be achieved, unless the company grows exponentially. (Which is improbable in this famously unprofitable industry.)
However, it's great they'll have a demo available, as this gives customers a realistic showcase of what they're paying for. It also proves that Mikoto has the basic pillars in place to produce an interesting engine, which is what most vocal synth fans will be invested in. Ultimately, most projects fail before they get this far, which is my primary source of optimism.
This demo comes alongside their crowdfund announcement. Particularly in light of Maghni AI, it's worth noting that significant features are almost always cut before the final release of a crowdfunded product (when they're fortunate enough to release), so donate with the understanding that no promise is guaranteed. Companies, especially when it's their first commercial release, cannot accurately anticipate the potential pitfalls they will face, and as a result almost always overestimate what they can achieve with the time/money.
Mikoto Studio lists plans to produce virtual instruments, audio effect plugins, music hardware, live performance features, and six languages: it's unlikely this will all be achieved, unless the company grows exponentially. (Which is improbable in this famously unprofitable industry.)
However, it's great they'll have a demo available, as this gives customers a realistic showcase of what they're paying for. It also proves that Mikoto has the basic pillars in place to produce an interesting engine, which is what most vocal synth fans will be invested in. Ultimately, most projects fail before they get this far, which is my primary source of optimism.
That's good! It was originally intended to launch at the beginning of the month, but was delayed due to bugs. I believe there are still bugs with the installer, so it might be best to hold off downloading for a few days.
Huh, this looks pretty interesting. I might have to try this out later, if only for the UTAU support. Do we know if this early access version has any of the AI singers, or is it just UTAU? I will admit the UI looks pretty interesting and the ability to not have to transfer files between the synthesizer and my DAW seems like a cool concept, but it seems like they're trying to do too many things at once and it's going to come out as a mess of half-baked features and languages that don't quite sound right. But all this is from just a look, I'll have to try it out for myself. The inclusion of Indonesian is certainly a surprise! That's a bit of a random language to add, but I'd imagine those who speak it are quite happy right now.
No AI singers in the early access version to my knowledge! It seems focused on trying out existing UTAUs.
Okay so, i just got this program.
First thing that happens after installing is that I get a completely white window. Reopen, still white. Wait a while and it finally takes me to a sign in page. Type in the info of the account I made, click next, bugs out and takes me back to the sign in. I reopened it and that seemed to fix it? Kept acting weird though.
Next, I get prompted to enter settings. Upon selecting my utau voices folder, the program crashes.
Reopened again. Program doesnt crash this time. Finally...
Click in the track to select a singer and there just... arent any. Despite loading my singers folder. Go to add singer, click the path I want and then next, and I get a white screen again. Go back to see if theres a problem with my singers, and the program crashes. Again.
After some trial and error, I finally have some loaded, but it should not have been this difficult. I know its early access, but come on.
Okay, now I insert a note and get a billion errors thrown at me. And the voicebank isn't making any sound. And theres autopitch I think?
How can you say this is a vocal synthesis engine that "just works" when the only thing ive had so far is it just... not doing that. How can you make one of your headlining features be this hard to use? Is it because I'm using a multipitch?
Also it labeled the vcv im using as vccv english. I fixed that, but nothing changed.
And their support site is down. Great.
First thing that happens after installing is that I get a completely white window. Reopen, still white. Wait a while and it finally takes me to a sign in page. Type in the info of the account I made, click next, bugs out and takes me back to the sign in. I reopened it and that seemed to fix it? Kept acting weird though.
Next, I get prompted to enter settings. Upon selecting my utau voices folder, the program crashes.
Reopened again. Program doesnt crash this time. Finally...
Click in the track to select a singer and there just... arent any. Despite loading my singers folder. Go to add singer, click the path I want and then next, and I get a white screen again. Go back to see if theres a problem with my singers, and the program crashes. Again.
After some trial and error, I finally have some loaded, but it should not have been this difficult. I know its early access, but come on.
Okay, now I insert a note and get a billion errors thrown at me. And the voicebank isn't making any sound. And theres autopitch I think?
How can you say this is a vocal synthesis engine that "just works" when the only thing ive had so far is it just... not doing that. How can you make one of your headlining features be this hard to use? Is it because I'm using a multipitch?
Also it labeled the vcv im using as vccv english. I fixed that, but nothing changed.
And their support site is down. Great.
for some reason i missed news of this one, and went to check out the demos on their channel. just happened to click on the release of Jinshi's new demo as it was premiering lol-- great song by nostraightanswer of all people. i like jinshi's tone but oof, he sounds really fuzzy.
i also checked out Sayaka's demo and oh boy... thats a rough one. i get that its such an early stage but she doesnt sound AI. and i typically try not to hate on producer's songs bc of how amatuer this fanbase is, but i do not think that song sounds good at all. again she has a nice tone and i see potential if they smooth it out. but they really shouldve held off until she sounded better-- at least Jinshi passes to me as sounding like an AI bank, its just that clarity that needs to improve
EDIT RE using the program's early access:
i downloaded early access just to get a peek and yeah lol, bugs galore. for some reason more than half the text has turned up as.... wingdings of some kind??? it loaded up and the singer tab has the option to browse some UTAU banks but you do have to go to the source and download it yourself. my own UTAU voice folder isn't being read either/the screen freezes when i try to load one from my own folder.
i also checked out Sayaka's demo and oh boy... thats a rough one. i get that its such an early stage but she doesnt sound AI. and i typically try not to hate on producer's songs bc of how amatuer this fanbase is, but i do not think that song sounds good at all. again she has a nice tone and i see potential if they smooth it out. but they really shouldve held off until she sounded better-- at least Jinshi passes to me as sounding like an AI bank, its just that clarity that needs to improve
EDIT RE using the program's early access:
i downloaded early access just to get a peek and yeah lol, bugs galore. for some reason more than half the text has turned up as.... wingdings of some kind??? it loaded up and the singer tab has the option to browse some UTAU banks but you do have to go to the source and download it yourself. my own UTAU voice folder isn't being read either/the screen freezes when i try to load one from my own folder.
Last edited:
Oh dear, looks like they launched their crowdfund much too early. Most people are forgiving of bugs in early access, but if the majority of users can't run basic functions, it will cause them to lose - not gain - faith in the software.
Up to this point Mikoto has been smart about their PR: not making any timeframe promises, keeping their development schedule private - all avoiding them the embarrassment of public delays, and keeping pressure off the devs. It's easy to get overexcited and jump the gun with public announcements, but it seemed like they'd learnt from the many other vocal synth projects that made that mistake. So without any public pressure, I wonder why they chose to launch the crowdfund with their demo in this state? Perhaps there was monetary pressure on them, or maybe they vastly underestimated the standards of the general public.
I speculate, but fan-Discords can at times be dangerously supportive and uncritical: a bubble that leaves developers unprepared for the cold shower that is true public scrutiny. I hope Mikoto can get the engine into a more usable state within the crowdfunding period. Fast bug-fixing would be a show of skill that helps make up for the blunder, giving the community some reassurance.
Up to this point Mikoto has been smart about their PR: not making any timeframe promises, keeping their development schedule private - all avoiding them the embarrassment of public delays, and keeping pressure off the devs. It's easy to get overexcited and jump the gun with public announcements, but it seemed like they'd learnt from the many other vocal synth projects that made that mistake. So without any public pressure, I wonder why they chose to launch the crowdfund with their demo in this state? Perhaps there was monetary pressure on them, or maybe they vastly underestimated the standards of the general public.
I speculate, but fan-Discords can at times be dangerously supportive and uncritical: a bubble that leaves developers unprepared for the cold shower that is true public scrutiny. I hope Mikoto can get the engine into a more usable state within the crowdfunding period. Fast bug-fixing would be a show of skill that helps make up for the blunder, giving the community some reassurance.
Last edited:
I just got the Early Access to work (heh), and I have... thoughts.
One, the UI. It's fine, I guess, but nothing special. I do like the chosen color pallate though. However, it was a bit difficult to navigate at times, but nothing I couldn't do on my own.
Two, installing singers. Worked well enough, but it wouldn't let me target my OpenUTAU singers folder, which was a little annoying, but I just installed two banks through their method and everything worked well.
Three, actual synthesis. Good lord, the synthesis. It is... not the best, to put it mildly. I used the voicebank Vlad V Act 2, an obscure one, but I like him. Anyways, he has pitches recorded at C3, G3, and I think E3. I laid down a simple Do scale, and uh. It was bad. Almost the whole time it was playing, it sounded like either it added a LOT of vocal fry or just ungodly amounts of engine noise. On further tests, it turned out to actually be engine noise. Again, C3-C4 is the exact range this bank was made for, and it sounded genuinely terrible. Higher notes sounded better, but not by much. I'd share samples, but I don;t know if I'm allowed to here.
Four, errors. SO. MANY. ERRORS. I can't import any files, or error. I can't switch active voicebanks, error. Sometimes if I try to put down notes, error. I can't render some notes, error. There are just so many, good lord. And not being able to import files most of the time sucks. There's more I forgot about here.
Five, imports. When they DO work, they don't, really. No note lyrics render correctly at ALL, they're either vowels or just A.
I literally can't use it like this.
One, the UI. It's fine, I guess, but nothing special. I do like the chosen color pallate though. However, it was a bit difficult to navigate at times, but nothing I couldn't do on my own.
Two, installing singers. Worked well enough, but it wouldn't let me target my OpenUTAU singers folder, which was a little annoying, but I just installed two banks through their method and everything worked well.
Three, actual synthesis. Good lord, the synthesis. It is... not the best, to put it mildly. I used the voicebank Vlad V Act 2, an obscure one, but I like him. Anyways, he has pitches recorded at C3, G3, and I think E3. I laid down a simple Do scale, and uh. It was bad. Almost the whole time it was playing, it sounded like either it added a LOT of vocal fry or just ungodly amounts of engine noise. On further tests, it turned out to actually be engine noise. Again, C3-C4 is the exact range this bank was made for, and it sounded genuinely terrible. Higher notes sounded better, but not by much. I'd share samples, but I don;t know if I'm allowed to here.
Four, errors. SO. MANY. ERRORS. I can't import any files, or error. I can't switch active voicebanks, error. Sometimes if I try to put down notes, error. I can't render some notes, error. There are just so many, good lord. And not being able to import files most of the time sucks. There's more I forgot about here.
Five, imports. When they DO work, they don't, really. No note lyrics render correctly at ALL, they're either vowels or just A.
I literally can't use it like this.
Hi, Mikoto Studio creator here! We've been hard at work to squash some of the most prominent usability issues that have popped up during the EAP rollout. While we have prepared for a lot of scenarios, unfortunately reality has shown us that there are always more bugs than a small team can account for upfront. We have been working around the clock to prepare an update with bugfixes, hopefully this will relieve the worst issues while we work to address the rest. Our apologies for any inconvenience caused by the current wave of bugs, we'll do better going forward!
We'd like to invite all of you to share your opinions, bug reports, and feedback via Feedback - ExpressiveLabs. This way, your reports show up in our issue tracker automatically, making it much easier for us to address every single one of them. We're very thankful for all the feedback we have already received, and we look forward to improving Mikoto Studio to the point where it can truly embody the vision we set out to create. We hope you'll tag along for the ride!
We'd like to invite all of you to share your opinions, bug reports, and feedback via Feedback - ExpressiveLabs. This way, your reports show up in our issue tracker automatically, making it much easier for us to address every single one of them. We're very thankful for all the feedback we have already received, and we look forward to improving Mikoto Studio to the point where it can truly embody the vision we set out to create. We hope you'll tag along for the ride!