If it’s her either they did a deal with Eclipsed Sounds/Dreamtronics or are doing things in a “different” way.
Puzzled.
1 Like
It sure sounds like it. It seems like this would be an EULA violation, unless there’s some technicality based on the lack of user control, the fact that the isolated vocals aren’t downloadable, and that they aren’t charging for people to use it… but they’re still allowing public access to a product that is only licensed to individual users, even if it is in a restricted manner.
3 Likes
I searched it, and this one sounds just like Kevin! 
Kevin?
Edit: So, I went on the site and it’s really fun, but they’re named as: Emily and Kevin.
以前、日本のユーザー達がSongRにsynthesizer vが使用されている事実に気が付き、問い合わせしていました。RIFFIT社はDreamtonic社と正式な契約を結んでおり、規約違反ではないらしい事が判明しました。下記のリンクの末尾に問い合わせの結果について書かれています。
誰でもブラウザで簡単にAI作曲。AIボーカルも入って1日5曲まで無料で作れるSongR BETA登場 | DTM 誰でもブラウザで簡単にAI作曲。AIボーカルも入って1日5曲まで無料で作れるSongR BETA登場 | DTM
1 Like
Is this their base web?
If this is as good as so called “AI” gets … ?
In my opinion the only thing to fear is how unmusical it is and how ‘naff’ the lyrics are!
I realise the thread is about the source of the voice banks but really? is anyone thinking the output of this site is musically useful???
So, it seems to be ok with Dreamtronics and Eclipse.
Said that…
I don’t know how SV users may take this thing. For sure i didn’t take it well.
When i pay a licence for a software (wich is a tool) is also to have “the privilege” (if you will) to have access to something not accessible for free to anyone else i’d like things to stay that way.
Using Solaria (or any other licensable voice) in their site instantly de-evaluate the voicebank and the licence investment anybody paid for it did.
How?
Well, how long will it take to be flooded by terrible AI made music with very recognizable Solaria and Kevin singing over it?
I paid for Solaria (and SV) and right now i don’t think i will use it anymore just not to have my music associated with awful AI composed tracks that, no doubt, will be used as background by youtubers in every new video.
At least licence to websites as songr unique and dedicated voices instead of ***** up people who has been supporting you until now.
I was planning to get a license to Saros when released but seeing how things are going i really doubt i ever will.
I think it’s not even as good as using SynthV lite, which is free.
You have no control over how they sing, so there’s mispronunciations, a flat and repetitive melody etc…
It seems like a nice introduction to SynthV (if you know about vocal synths), but you can’t choose the music other than the 4 genre selections. It’s quite monotonous, but fun if you want to dabble in it quickly.
You literally just select a genre out of:
Pop, cafè, hip hop and piano rock - prompt the lyrics, which you can edit, and when you’re finished, you render it. It comes out how it comes out - that’s it!
You can’t change anything to it like you can in SynthV, so don’t feel disheartened. You can only generate songs 5 times a day too; so it’s limited.
you are right but this is the actual situation.
I’m sure they plan to expand it way further.
1 Like
Like maybe actually creating a tune instead of that garbage, or lyrics instead of meaningless half-sentences carelessly bolted together?
Actually, that reminds me of a BASIC program I had on my Sinclair Spectrum (1980’s cheap home computer) that tried to string phrases together into short stories - it was about as successful (or bad).
1 Like
I’m sure they’ll constantly strive to improve it, but I think it’s probably seen as an extension to SynthV Lite, and probably won’t stay free either?
There’s absolutely no way it will ever compare to SynthV AI, and I still plan on buying Solaria AI too!
Completely awful and amateurish. This use of AI has VERY far to go.
If people are stupid enough to imagine that buying a few pieces of software replaces musicianship and natural talent then they deserve all they get.
As to the argument about the voice databases - that I suppose is up to the makers as to whether they will allow this use?
A common misunderstanding is quite well explained here:
2 Likes
Looks like it is Solaria, but in my opinion to make Solaria sounds realy natural you need to tweak it and spend some hours on pitch variation, breath etc… So, this kind of automaticaly generated stuff will never be as good as a human can do with SynthV. Or it may be one day, but then, you’ll have to prompt some intention to direct the singer…
I’m still wondering if giving acces to music creation to anyone is really a good thing? I mean, to me, art is expressing oneself soul. Using automaticalli generated AI stuff, doesn’t really allow for self expression… What do you think?
Cheers,
That’s the part of this that I find almost comical. Their website says “Our mission is to democratize the creation of song and music for all”, but… free music creation software already exists. Anyone can get a trial of Reaper or download Cakewalk and make a simple tune. There are plenty of easy-to-use mobile apps for creating various types of instrumental sequences.
Music creation is already available to anyone who wants to create.
It will surely take more time without the prior knowledge, but even pure trial-and-error still allows the user to engage with a creative process with no understanding of music theory. If someone struggles to write lyics, purely-instrumental genres exist (or they could just practice the skill like people always have…), and if someone can’t sing, free vocal synths exist.
None of these things are beyond simple study and practice.
Even if SongR were to advance to the point of creating professional-quality tracks with a click of a button (doubtful), I have to wonder why? Would that really get listens on Spotify? Would it offer users anything more than a brief moment of novelty?
Doing the work for someone isn’t enabling them to be creative. Even in a hypothetical situation where a user could direct the AI process in a pseudo-conversational manner… we don’t say that a director is “acting” when they instruct actors on a movie set, why would we treat instruction of a music generation AI any differently?
3 Likes
Totally agree…
Google translate of the link Jack_O_Cherry posted for anyone wondering:
*Added on May 14, 2023
The singing voices were Kevin and Solaria from Synthesizer V! !
When I saw the name Kevin, I thought, “It’s the same as Synthesizer V…”, but many people on Twitter pointed out, “This must be Kevin’s voice from Synthesizer V.” It’s true that their voices have a similar atmosphere, but on the other hand, I thought “Emily isn’t in Synthesizer V…”. On the other hand, some people pointed out on Twitter that “Emily is probably the voice of Solaria!” Solaria is a singing voice database for Synthesizer V by Eclipsed Sounds , which was introduced in the article " [NAMM2023 Report 3] The World of Singing Synthesis Aimed at by American Synthesizer V Vendor Eclipsed Sounds " the other day. In order to find out the truth, I contacted the developer, RIFFIT, via Twitter and email, but as it was the weekend, I have yet to receive a reply. On the other hand, when I contacted Kanru Hua ( @khuasw ) of Dreamtonics , the developer of Synthesizer V , I immediately found out. Kanru, who was just on a business trip to Berlin’s Superbooth, said, "Actually, we’ve been communicating with RIFFIT for a long time. SongR has just been released, but it’s already become a hot topic!’ said a surprised voice. It looks like it was Kevin and Solaria, exactly as you pointed out. " I was thinking of announcing that it was Synthesizer V’s voice when the quality of the composition was a little higher, but… " Some people may have been worried about the unauthorized use of Synthesizer V, but it seems that they had signed a license agreement with Dreamtonics before releasing it. I will let you know if there is an update of the information.
*2023.5.15 postscript
Email reply from RIFFIT
This morning, I received an email from Dr. Lex Van der Ploeg , CEO of RIFFIT . As Kanru told me, he has been working closely with Dreamtonics on development, and is using Synthesizer V’s Kevin and Solaria. Also, since I would like to inform the readers of DTM Station, I received some information in bulleted form, so I will post the translated version as it is.
- We have several unique joint software packages for the functionality of SongR and RIFFIT Reader ( https://www.riffit.com/ ).
・SongR is for general use. The RIFFIT Reader is intended to improve individual learning/reading comprehension and to be used in the classroom.
The use of the SongR/RIFFIT Reader to treat people with learning disabilities and to improve communication in general is supported by issued US patents.
A wide range of applications for communication and entertainment are spreading rapidly.
• These actually have several audio libraries, including Solaria, which he uses for Emily, New York Broadway, his artist Kevin, and others.
・We are working closely with Dreamtonics and are leveraging Synthesizer V.
I personally don’t worry too much. Every AI Music website I’ve seen so far is made of comissioned music pieces with slight variations in the essentially same piece, think of it like 10sec. music clips for videogames but more monotonous and without engineering. Although Nvidia presented its new GPUs, they have been caught lying in the past and at some point the power bill just becomes unaffordable which is why companies look into analogue computing without any general progress worth mentioning. Dreamtonics is probably just taking money from whoever gives it to them in this case.
1 Like
As Andy said ‘If there was a button saying Hit Single, I would have pressed it’ The one I got out of this site (using some AI generated lyrics i got from ‘These Lyrics Don’t Exist’ worked really rather well. I have listened to it several times now. I’d say the lyrics were no worse than most modern pop music.
What I’d want this site to do is let you download the midi (including lyric midi for the vox) so I could work it into something much better.
This is a toy but someone could do something similar (with sensible parts downloads) and it would be quite a useful tool.
Possibly not quite the Hit Single button though. Machineguns and trains - YouTube
As to the argument about the voice databases - that I suppose is up to the makers as to whether they will allow this use?
Absolutely.
As is a free choice of a customer to continue to support or not a company that from a day to the other made the customer investment in their product(s) ways less valuable.
Andy McClusky (OMD) explains
My arguments were never about this.
I appreciate that, dhstsw 
I only mention it because it is often-used by disparagers of those who are successful in the Music Business 
At this current stage, “original” music from AI still sounds very, very bad–almost like just random notes. Google’s AI Test Kitchen is just as bad at creating music composition.
But, I have no doubt that eventually, AI will become pretty good at creating original music. I don’t know how long it will take–maybe 10, maybe 20, maybe 50 years–but I’m sure it will get there at some point.