Other SymphOneme Discussion thread

[pico]

It almost reminds me of a worse, pitchy version of CeVIO.

No matter what your opinions on CeVIO are, this is being too mean to CeVIO

In all seriousness, uh... yeah, it sure is! I wonder what the target audience of this was supposed to be. Maybe it really is a passion project? Is there corporate backing for this? Very mysterious

 

[REDD]

No matter what your opinions on CeVIO are, this is being too mean to CeVIO

In all seriousness, uh... yeah, it sure is! I wonder what the target audience of this was supposed to be. Maybe it really is a passion project? Is there corporate backing for this? Very mysterious

A part of me does suspect this is a passion project by a small team. While im not exactly an expert in software development (please correct me if i say anything inaccurate), IIRC a beta version of a software is supposed to be at least somewhat useable? I mean, it certainly FUNCTIONS... but the result is almost unlistenable. While i like the tone of some of the voices, (i like Mel as a concept) the pitch is inconsistent at best, and phonemes are hard to hear sometimes.

I highly doubt this is a project by a larger company due to how unusual the whole concept is. While multimedia projects aren't a new thing, typically when you make a lore to go with a game/program/whatever you typically have both mediums ready at the same time. (to my knowledge) IIRC, theres no manga/anime/game series with these characters even ANNOUNCED at this point, which would suggest that these are just some guys characters from a story hes been trying to write. Going up to a big corporation with this idea would be a hard sell.

The characters are nice, dont get me wrong, and if executed well, these voice types could really shine, but unless they SIGNIFICANTLY improve the quality by the time this is released, this isn't gonna go anywhere. Sorry if i'm coming across as harsh lol

 

[peaches2217]

I can get in the website, but for some reason it's incredibly buggy and keeps disappearing my files. I tried a few different things with no results so I gave up and looked on Twitter for someone who already used them Also the demo buttons on the main site work now so if you want to listen to them you can.

Mel

Rasetsu

Enemoprem

Look, I honestly don't know what I was expecting, but my God, it wasn't "2008 monopitch CV Utau recorded on an iPhone 3 by an 11-year-old who just downloaded Audacity last week". This isn't just bad. This is the single most beautiful and glorious disaster I've seen in a good while.

 

[Mika]

Look, I honestly don't know what I was expecting, but my God, it wasn't "2008 monopitch CV Utau recorded on an iPhone 3 by an 11-year-old who just downloaded Audacity last week". This isn't just bad. This is the single most beautiful and glorious disaster I've seen in a good while.

Nonono, I've been around for 10+ years in this community and every single badly recorded monopitch CV Utau I've heard is much better than whatever this is lol.



Yeah this is just sad. The concept was interesting and the voices seem to be chosen well but man it just sounds so awful. With a ton of work and much more time this could be kinda good but.. I think they shot themselves in the foot. WHO wants to support them now with all these restrictions and then you get this kinda sound. Yeah, no.

 

[Tortoiseshel]

At first I wanted to keep an eye on this just because it was a new synth that seemed to be doing something different. Now I want to keep an eye on it just to see how bad it can get.

 

[uncreepy]

At first I wanted to keep an eye on this just because it was a new synth that seemed to be doing something different. Now I want to keep an eye on it just to see how bad it can get.

I'm sure it can only improve, it would be weird if they stopped working on their software and teaching the AI like "We're done!". (So the quality would either stay the same or get better, not randomly get worse haha) Maybe they have gotten some feedback from Japanese users about the quality or their rules, so they would have things to consider changing.

 

[ikaroll]

Me: oh hey there's a new synth, I'm gonna check out the thread!
Me, scrolling all the way to page 3:

All I can say is....yeah...it sure does sound like a singing robot computer program

 

[PearlStarLight5]

So I'm just listening to these demos now and
they called it way too early to put it on beta, like how did they release this without expecting to make a total laughingstock of themselves?
Also that Rasetsu sounded the worst of them, whyyyyy
In conclusion, Miis are better singers than them

 

[MillyAqualine]

Also I don't know if we can say it yet (though some prooves tend to show up more, directly into the AmadeusY servor) but... Let's say I start getting very bad vibes about that project, without mentioning the quality of the voicebanks

 

[CdrSonan]

I can't really comment on the allegations regarding AmadeusY, but I also agree the program appears to be in a rather poor state.
A Beta is normally used for polishing, fixing issues, implementing feedback, and optionally some small new features as long as they don't break compatibility with previous versions.
Anything that does so, and all large features in general, should be implemented during the Alpha stage.

I've written about this on Discord before, but I still wonder what they did with their AI. Assuming they use an NNSVS-like synthesis pipeline, they are going to have three separate AIs (timing, pitch, and final sound), and the program produces errors with all of them.
However, this isn't really typical for an AI that wasn't trained as much, since that gives you (engine) noise, not errors. So I think it's possible they're dealing with an AI that's not well configured in some way.
The process for finding the best configuration is called Hyperparameter tuning, and it needs even more compute power than training. It either takes really long if you do it on a PC (if that's feasible at all), or it's expensive if you rent a server for it. It also means more programming work.

So my theory is that this really is a passion project, they wanted to save time and guessed which configuration might work. (This can often work btw, but it's always a gamble)
But if the issue really is AI configuration, this would be extremely bad for them, since you have to restart AI training from basically zero after a configuration change.

 

[aru ii]

There is now an account called Symphoneme Law and it's followed by Symphoneme's account... did they make it to expose people who make fan art of their characters or what lol

https://twitter.com/SymphonemeLaw

EDIT:
One of the devs (I think it's one of the devs?) of symphoneme said that they are working on improving it and that theyre thinking of implementing a browser based gui to imput the notes back in november

 

[mary34]

It migth be in response to the allegations made by one of the Amadeusy devs about code stealing? (It's probably not that but who knows)

I am more shocked by the fact that this project is still alive and kicking...

Update: Yeah it's def something to do with AmadeusY. I went to check the account and their only tweet so far has them @ the AmadeusY account asking them to contact their legal team if there's any issue via e-mail. (Full disclosure I used machine translation to parse what the tweet said so take this with a grain of salt.)

Another quick edit, but I just got home and noticed the date is from last year so this might be old news at this point.

 

[aru ii]

Symphoneme is alive again. The free beta will end on September 27 and later the full (probably payed?) version will be released later. Maybe the 2 new singers will also release with the full version?

 

[pico]

> 本日が開始から1年ということで
> 1年
終わった…

It’s already been a year?! Time seriously flies…