… then I think it’s very important/valuable to watch Benn Jordan’s recent video on the technology and licensing model for http://voice-swap.ai
It’s probably good for most people, but I don’t like what I hear. To me, it sounds flat, unnatural, and on-off-ish - Just like much of today’s vocals are often beaten and tamed to death with compression and autotune.
Not directly related to the Video but on the subject of AI art…
I like my AI curing cancer and analyzing malaria data… and if it wants to recommend new songs to me on Spotify I can live with that… beyond that as far as it being the (only) artist on the track it doesn’t interest me at all.
As we hurtle so stupidly into technology that we can’t control or even comprehend we have probably already raced past that point on the graph where the balance point was that it is actually beneficial to humans mental development to take at least some time, effort and sustained attention to developing artistic skills and from that point then employing technology to streamline the manifestation from idea to work of art. Of course since the advent of wax cylinder recording we’ve been on a sliding scale of technology and art that of course each generation has enhanced but as with any scale balance is the ideal.
Will people have any sort of lasting satiety with this sonic aspartame? Will music without spirit or soul have real appeal to spiritual beings with souls/intellects? Perfect autotune has become a meme, will this as well?
I didn’t post it because I thought it was great sonically. I don’t think Benn thinks it is amazing either. My interpretation of his point(s):
- the quality improves when you actually have high quality recordings
- you get high quality recordings by actually paying people to make them
- the results are much better, and it seems likely that they are already or soon will be usable by musicians with no realistic budget for hiring a singer
- the concept offers a chance to think about licensing fees and structures, and we need to do this early this time, given the way musicians totally missed the boat (or were kicked off it) for the streaming revolution.
When I listen to the Hindi version of Amazing Grace in the middle of the video, I do not find myself thinking “this has no spirit or soul”, though that could be because I don’t speak or understand Hindi, nor am I familiar with the inflection styles when singing in Hindi.
I find timbral transfer and voice replacement two areas of potential “AI in music” that seem to me to brighten, not darken, musical creativity. They have almost nothing to do with generating compositional ideas, and everything to do with giving a musician access to a sonic palette that would otherwise remain inaccessible.
The downside, as with DAWs, is the diminution of music as a social activity and music as a social process.
I agree with your interpretation. I managed to only go to http://voice-swap.ai and write from what I saw and heard there and missed Jordan’s video. As such, I went off-topic, sorry about that. After listening and looking at that video, I really dislike it even more, but only what I hear. I like that more people get the opportunity to create and make music in new ways.
…and so in the year 2050, the mighty spaceships hovered over the great capitals; Beijing, Sao Paolo, and the Eastern Sprawl.
The communication came through:
“We come from the transgalactic federation of Zoog. If you wish to join, and enrich yourselves in our splendour, all you need to do is give us your culture, show us your great artists!”
“Ahhh, well…all the artists are gone. We sampled a few of the really rich ones - or at least the ones with rich daddies who could pay lots of money under the table to stomp on the competition. All the really talented people eventually gave up, so now we totally use computers and AI to make all our art now”, replied the Earthers. “It’s about equity. No talent necessary! Look, we can even make puppies sing!”
The great alien leader looked bewildered.
“What? Why would you want to do that? You were naturally endowed with great talents for art! It was a solved problem! What happened to the cool stuff that was on your Voyager probes? You replaced them with AI? Weirdos.”
“No…they never passed the latest algorithm. I mean, the people liked them, but, well, they eventually got ranked so low that noone really knows that stuff. Hehe funny puppy!”
The emissary sighed.
“We already have AI and computers - but that’s maths, not art…there’s no inspiration, no perspiration, just sitting in a chair clicking buttons and trying to fit in with an algorithm…well, I suppose you have nothing to offer the universe, and therefore we rescind the invitation. Sorry about the mix-up! Cheerio!”
So the massive spaceships left for interstellar space, never to return.
And then there is hatsune miku and the whole vocaloid movement, with “liveshows” etc. since about a decade or so. Humans will gather around any fire shiny enough.
Edit: In other words, realism does not matter. We look disney animatino and know full well it is not real, still, there is emotion. Fully automation created content will find it’s consumers. It is just a question for the creator if they want to cater to such an audience or to those who seek a connection to skilled artists telling a real story.
THIS, THIS, THIS!
At time of posting, there is a major strike by some of the main unions in the Entertainment industry, as scriptwriters and actors are, rightly, concerned that large industrialised media conglomerates will use AI derived versions of their scripts and images instead of hiring them.
The current laws are not fit for purpose to guard against this, and the large media corps know this and are lobbying to keep the status quo so that they can exploit it.
There are strong parallels with the music industry here.
I think it’s a key time for artists and creatives to harness technology and set the precedents for how they are treated and compensated in the future.
And the initiative Ben talks about is a good start, but we need more initiatives like this across the industry.
A local friend of mine uses Hatsune Miku to provide vocals for his guitar performances. It’s not my thing, but it’s a particular style and very distinctive, unlike AI which is trying to make you sound like a real artist.
I suspect he will not mind me posting an example of his work, for anyone interested.
It is good that you mention the aim of realism. The vocaloids are also moving into that direction, but back when it started, tech was not advanced enough. Now there are competitors who achieve impressive results with phoneme libraries combined with trained tech:
But I have to apologize that I was unclear about my point mentioning Hatsune Miku. It was about the question if that kind of creation can be perceived as having a soul, mentioned earlier. Sorry for being unclear on that. I am a bit all over the place sometimes
Well, if you understand Polish, there is an presentation of polish algorithm of AI voice swaping. There are no subtitles, unfortunately. At the beginning Monica speaks with voice of David, icluding intonation, etc. I think the effect is very convincing.
26:15 - AI modelled voice in a song (short sample)
That’s a great piece of video (Lucy’s “voice” is wonderful, lol).
I welcome AI in music production.
Does it sound natural? Probably not in most cases. But how natural is it to double-track a vocal? How natural is the “choir” in 10cc’s “I’m Not In Love”? The thing is: it sounds great and expresses emotion. It makes the listener feel something different.
I don’t go for using AI to create a song. Not for puristic reasons. Just because it’s not fun. Part of the fun is the freedom to create yourself.
But I totally dig AI in mixing and mastering, for example. I’m spending months trying to learn to put the sound right because I’m not a mixing engineer, just a musician. It’s just not viable for me to pay to mix and master each record I want to produce. So if there was an AI tool to give me what I want than I’m totally fine with it.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.