Voice attribution for lipsync music video

Of course to make it really awkward, although I can’t sing (throat issues) I can put a tune in, use something like waves tune, melodyne or JamOrigin midibass to get the midi, which I then put the words in in Synth V. But I think at this point it’s getting to sophisty so there’s no need for a reply to that.
However I’m working on a track at the moment where the main vocal is a male-fied version of Solaria, tuned down an octave in Soundforge then some filtering over the same parts done by Asterian and Solaria to give a single voice - how would I credit that?

1 Like

Oh yeah that’s an interesting one, and in some cases a bit of a grey area (and also a cool idea, I hope it works out well for you!).

I think your situation is simpler, but I’ll first go over the potential pitfalls and complications:

For example, this is an official vocal synth version of a song by Tacitly, made by KIRA. Notably, the vocal synth version is not available for sale or on streaming platforms like Spotify (perhaps because of this grey area issue).

This song uses Hatsune Miku and GUMI (Vocaloid products) as the main vocals, however people familiar with vocal synths noticed they sounded an awful lot like Tsurumaki Maki and Stardust Infinity (SynthV voice databases) in certain places.

So while I can’t say this for certain, it seems like a similar technique was used (perhaps using Zynaptiq Morph 2 or similar) to combine the two vocals, with the goal of maintaining the qualities of the Vocaloid synths with the added realism of SynthV.

It’s possible if this song were to be sold commercially they would have to specify Maki and Stardust as supporting vocals or similar in the metadata/description, but it’s a new enough technique that we don’t really have precident for it (and it’s also important to note this is combining two competing products).

(edit: one thing I initially neglected to mention is that Maki and Stardust do not have the same license terms as the first-party Dreamtonics ones, so their requirements may vary somewhat, but the original statement remains true if we were to apply the situation to the Dreamtonics products)

As for your situation with Solaria and Asterian, I think you have a few options:

  • Credit both equally
  • Don’t credit either
  • Credit the one it sounds more like as the vocalist, and the other as backing/supporting vocals
  • Reach out to Eclipsed Sounds for clarity ( [email protected])

You can also see Eclipsed Sounds’ statement on the use of their products as the input to voice changer models such as the Vocaloid6 Vocalochanger (keep in mind that statements by Eclipsed Sounds to not affect terms for voice databases other than Solaria, Asterian, and potential future products by them specifically):

This part is especially relevant:

While it is not required if using the full version of the voice database, we would greatly appreciate if users who utilize exported audio for this purpose would also credit our voice databases for the initial rendition in order to acknowledge our team’s hard work on our vocals.

1 Like

haha dude @Chris, if you like Vocaloid, use it! Whether that’s for the program itself, or the company’s wording of their policies.

Sounds to me like Dreamtonics is just covering their bases. Do they really care if I do this? Probably not. But if Disney makes a new musical, using SakiAI to voice the main character, but credit the singing to a different person/character? That’s probably why they have a “policy” in place.

How about the case of a violin VST - let’s say, Joshua Bell violin. I could use that VST without mentioning it by name. OR I could say “created with Joshua Bell Violin”. But I cannot say “Violin played live by Andre Beller” or “Violin created with Yamaha DX-7”


Great discussion! This topic really is going to become a ‘thing’ as AI develops. I’m old enough to remember similar debates when sampling was a new thing…and even further back, when drum machines became widely available.

Btw, I contacted Eclipsed Sounds, for SOLARIA; I just wanted to check if what I wanted to do was okay. Their license is clear on the restrictions on how SOLARIA should be credited, but it’s allowable to give no vocal credit at all, which is what I’m doing. They were extremely helpful :slight_smile:

IMO, we’re at the point where Synth V is an amazing tool. Its AI component is limited to minor voice articulations. It doesn’t write the lyrics for you, it doesn’t write the melody, and it doesn’t construct the song. I feel I’m in complete creative control. When that stops…maybe I’ll take singing lessons!

1 Like

For sure. For the time being we have 15+ years of precident for commercially successful vocal synth music, but that won’t always be the case and things may change.

Additionally, vocal synth music has only had limited success breaking into English markets in that 15+ year history. There’s always the possibility that a different approach is necessary for a different market, but I do think it’s valuable to be aware (at least at a basic level) of the history in the Japanese/Chinese markets where it has become a somewhat normalized category of music that can stand on its own.

Despite the AI aspect, SynthV Studio really doesn’t play any different of a role in creation than the first successful Vocaloid products did.

The main differences are:

  • Historically the character/name of the vocal synth has been considered a beneficial part of the product, so people felt no need to hide that they were using a synth
    • It has actually been shown to be preferable to acknowledge which vocal synth is used for a song, because it drives discoverability and provides that audience recognition/“story” mentioned by @pderbidge above (for example, your song will get more traffic if you label it with “ft. Solaria” simply because people will search for songs using Solaria and discover your music that way)
  • SynthV is finally capable of producing high enough quality output that listeners might mistake it for a human vocalist, which hasn’t really been the case with other synths like Vocaloid or UTAU (which is probably what prompted that specific line of the Dreamtonics license)

It’ll certainly get more nuanced if/when the software starts to encroach on more of the creative process though.


Blockquote […] your song will get more traffic if you label it with “ft. Solaria” simply because people will search for songs using Solaria and discover your music that way)

Ah, yes, I hadn’t considered that. Makes sense. And that background info is useful to know, thanks.

Thanks for that.
If you’re interested I have put up When Does Protest Become Simply Whining | Dored | DOD where track 3 (lying on the Floor) is the track with the manipulated/combined vocal.