December 1, 2024. I tried to hate Christmas really hard, but I gave up. For some reason I am sensing hints and pixels of a 2014 Christmas vibe setting on early and lasting into late winter next year. There is something about that time of innocence, naiveté and optimism from 10 years ago that appears to be quietly returning. It is a very counter-intuitive vibe (in light of everything that is going on) but it appears to show up independently from whatever is happening.

This also means that 11 years ago, it was just as independent from world events as it is now, but we used to attribute our bright feelings to the blind trust we had in our world leaders, our institutions, our elect-representatives, and our justice systems. Although literally everything in the realm of trust has collapsed since then, the non specific vibing mood appears to be seeping back on. Lightness of being is remarkably more bearable than the other version.

These are from my kpop dance practice series. I am heavily influenced by Kpop both in music and dance. These choreographies are usually performed in typical combat boots (which I love btw). Not today. Here, I'm doing the chorus of Chk Chk Boom by Stray Kids twice at 100%, the chorus of Untouchable by Itzy, once at 90% and 3 x the chorus of Miovv by Miovv at 100%. These songs were all released in 2024 and dance practice videos of the full songs are always available online.

Dance happens to be the only thing AI can't do at all, so humans continue having a significant advantage in that regard. Kpop is, in that specific window of relevance, likely, solving the greatest existential crisis we've ever faced as human beings. If you can follow the moves of a kpop chorus, first in your head, then with your body, you can convince yourself with certainty that you are not a bot. If anyone ever accuses you of being a bot, respond with a kpop chorus routine.

---

November 16, 2024. After making a ton of new sound, as usual I am unable to continue "liking" the songs I made after I got to listen to them over and over during tweaking. There comes a moment when I've worked so much on the same song, I'm just done with it, and I want to make something completely different. It is becoming ever more clear that none of these songs will be released during my lifetime. I care more about creating new content for now. I'm also falling back on traditional music tools such as Splice samples. So here, I veered into a completely new territory producing rap songs from my favorite sections in the Criminal Code. I start verbatim with ss. 273.1+273.2 in French. This is something that can sustain a kpop like dance routine.

Before the advent of AI, I used to make vocal recordings of the entire code in French and English that I used to play in my sleep before exams. It fixed the code in my head, helping me to save seconds during time-intensive multiple choice answers. So here I go with the same section in English, with a slightly different sound.

As of now, only a small percentage of AI music is public, many people are afraid to post their creations due to lawsuit threats from everywhere with little action by the courts. As you can see, one way to completely go around that, is to use not-anyone's-lyrics in AI prod, you first need to understand all the music genres and see how they mix together. Here is a rap of s 184, Interception of Private Communication, again in French. Male vocalist this time.

In Suno you can enter prompts up to 5000 words. A typical section from the code averages about 4500 words. I personally don't post on Suno, but for those who do, know that the "reuse prompt" command in Suno allows you to use someone else's lyrics and put your own music or vocals to them, or rewrite or translate the whole thing and completely make it your own.

---

June 8, 2024. I released another 6 tracks and added to the same album in progress. Two of them are using identical lyrics. The fact that there is no issue with musical rights across platforms means that artificial intelligence is trained to work around existing copyright, the same way DJ producers do it, by taking de minimis samples from everywhere and making it impossible to get yourself sued. There was a certain ambiguity with respect to ownership in AI compositions, but it is resolved, for now all tracks are copyrightable to the human user (not the AI platforms), so long as they contain original lyrics or lyrics generated from original lyrics. Evidently, if you use other people's texts you are on the hook.

The difficulty these days is not to get the music done, but the fact that there are infinite possibilities of songs that can be instantly generated from the same lyrics and you can lose yourself in the labyrinths of sound before you set your mind on one particular version. It requires discipline. All of these versions automatically receive protection under the Copyright Act as original musical compositions.

Vocal interpretations are supposed to be protected under section 15.1 et seq. but we all know the ire I attracted when attempting to assert that section in 2018. The Quebec Court of Appeal said that unauthorized vocal recordings used in a musical work are not protected if the vocals are not a musical work, so I could only assert a privacy claim (intrusion upon seclusion equivalent) for the vocals. This is not what section 15.1 says, but the courts in Quebec, as my civpro prof so eloquently said, are like a casino. If you are not into gambling, it is best to avoid them.

Since privacy rights are NOT considered economic per se, the damages are super minimal and cap at $5K in most provinces (the most we've seen is $10K for breaching someone's bank account in Ontario) you can't even cover your execution costs with such amounts so it is a guarantee that no lawyer in this country will ever take a case involving unauthorized use of cloned vocals in Canada. Canadian law firms barely make ends meet, so they can't afford the same generosity as their US colleagues where the private litigation sector (especially in IP) is super healthy. In the US, unauthorized vocal cloning is also litigated as a privacy right (appropriation of likeness), but damage awards are higher and judges work very fast (in comparison), with the added benefit of civil jury trials where damages can skyrocket. So, the US is obviously a better forum for IP litigation, BUT if someone wants to continue cloning your voice in Canada, there is nothing currently in place that will stop them. So, unless the whole world bans cloning of vocals, Canada will always be a weak link in enforcement.

The other problem I see with appropriation of likeness, taking it from the Bette Midler case, is that US law seems to distinguish between vocals of someone famous and vocals of someone less famous. I find that feudalistic and arbitrary. Does the law apply to everyone the same way or does it distinguish between industry-backed plaintiffs and ordinary artists minding their own business while commercially exploited without their consent? The LOVO class action will answer that question because it mixes famous actors, less famous ones, and ordinary narrators whose voices are currently up for grabs.

I am not sure the class action will be certified, precisely due to the feudalistic interpretation of privacy rights in US precedent. I really don't have a very high opinion of US privacy law. That Bette Midler precedent is problematic. Voice cloning is a grey zone and I think the only way to make it count is to create a new criminal offense and make it enforceable worldwide.