Musician and Data Scientist. Founder of Infinite Album & Melodrive -Creator of Ditty - on a mission to enable musical creativity through technology.

Joined April 2009
56 Photos and videos
Pinned Tweet
We've fully launched 🥳 Had the amazing opportunity to work with @ThomasBarrandon to launch our first Artist Pack - with music infinitely generated in his style.
IT'S FINALLY HERE! Unlimited streams of copyright-safe AND game-reactive music for your livestreams! Infinite Album uses fully licensed generative AI music to create infinitely playable streams of music. #twitch #twitchstreamer #twitchaffiliate #twitchpartner #TwitchconVegas
8
808
Trying to bring a bit of transparency to AI in the song making process. I spearheaded this project to open-source our submissions at the AI Song Contest, and our dataset paper just got through ISMIR’s peer review 💪
The AI Song Contest is proud to announce that our dataset paper has been accepted to @ISMIRConf 2024! We are making our 2023 submissions public, to encourage #transparency in the #AI & human co-creation process. We worked with @UW. Attached is a small sample of our analysis!
1
4
344
Suno is getting sued. I feel like music has one of the more restrictive legal frameworks, and they’ve basically been through this before with Napster. It’ll be very interesting to see how this plays out - could be a power play to position a good deal w/suno.
🚨BREAKING: @UMG, @capitolrecords, @sonymusic, @warnermusic, @AtlanticRecords, @Rhino_Records & other major record labels sue AI music program Suno AI for copyright infringement. Important quotes: "Given that the foundation of its business has been to exploit copyrighted sound recordings without permission, Suno has been deliberately evasive about what exactly it has copied. This is unsurprising. After all, to answer that question honestly would be to admit willful copyright infringement on an almost unimaginable scale. Suno’s executives instead speak publicly in exceedingly general terms. For example, one of Suno’s co-founders posted online that Suno’s service trains on a “mix of proprietary and public data,” while another co-founder stated that Suno’s training practices are “fairly in line with what other people are doing.” Piercing the veil of secrecy, an early investor admitted that “if [Suno] had deals with labels when this company got started, I probably wouldn’t have invested in it. I think that they needed to make this product without the constraints." (page 3) - "Plaintiffs could have proceeded with this action based solely on eliciting that reasonable inference of copying. Nevertheless, Plaintiffs’ claims are based on much more. In particular, Plaintiffs tested Suno’s product and generated outputs using a series of prompts that pinpoint a particular sound recording by referencing specific subject matter, genre, artist, instruments, vocal style, and the like. Suno’s service repeatedly generated outputs that closely matched the targeted copyrighted sound recording, which means that Suno copied those copyrighted sound recordings to include in its training data. In addition, the public has observed (and Plaintiffs have confirmed) that even less targeted prompts can cause Suno’s product to generate outputs that resemble specific recording artists and specific copyrighted recordings. Such outputs are clear evidence that Suno trained its model on Plaintiffs’ copyrighted sound recordings." (page 5) - "At its core, this case is about ensuring that copyright continues to incentivize human invention and imagination, as it has for centuries. Achieving this end does not require stunting technological innovation, but it does require that Suno adhere to copyright law and respect the creators whose works allow it to function in the first place. Plaintiffs bring this action seeking an injunction and damages commensurate with the scope of Suno’s massive and ongoing infringement. (page 6) ➡️Read the full lawsuit below. ➡️To stay up to date with the latest developments in AI policy, regulation, and lawsuits, subscribe to my newsletter (link below)
2
249
Ryan Groves retweeted
I don't think people elsewhere fully grasp what's happening in NCR and just how hot it is. At 7am, the tap water is boiling hot. The sun hurts the eye. There's no night time anymore. For 24 hours the temperature feels above 40C. Which means during night the water is as hot
423
2,982
13,457
1,545,990
Ryan Groves retweeted
Questions at academic conferences:
105
1,221
9,165
998,739
Ryan Groves retweeted
94
2,982
29,723
1,969,026
Ryan Groves retweeted
First Adobe changed their Terms to give themselves a permanent license to any content you produce using their software. Then they pretended it was a misunderstanding. Then Adobe gaslit people who pointed out that they are full of BS. Then they put out damage limitation statements that mean nothing. I have cancelled Adobe and so should you if you are a creator who has a backbone. Here is what I have done: 1. Premiere Pro -> Davinci Resolve This is an upgrade. The software is less clunky and actually works better. Integration with hardware is better. No licences. No monthly payments. I bought the Speed Editor Keyboard that comes with a Resolve license - absolute bargain of a deal. 2. Photoshop -> Affinity Photo This is also an upgrade. After an hour of getting used to it, Affinity software actually works better. Image editing and my YouTube workflow are better. 3. Illustrator -> Affinity Designer Same as above - actually very good software. *** In the last few years Adobe’s competitors have caught up with and overtaken Adobe. I was blind to how good the alternatives have become, using Adobe stuff because I was a creature of habit. Would very strongly recommend trying the alternatives out. You might surprise yourself… And save a boatload of cash.
476
3,328
13,451
1,270,032
Ryan Groves retweeted
An important read regarding Cara and the reality of large scale operation cost. Angelo makes some great points here and has a lot of experience in this area. Jingna and her team are already facing massive operational costs and Cara is still growing.
87
444
2,147
289,285
This needs to be shared as far and wide as possible so that no one ever has to code formulas in latex ever again 😅
Microsoft Copilot just blew my mind. I wanted to see if it could generate LaTeX code from an image, and not only did it do a fantastic job (almost 100% correct), but it added some context about the formula itself! 🤯 I'm not coding LaTeX formulas anymore, this is a game changer
1
110
Ryan Groves retweeted
More proof that generative AI competes with the data it's trained on. 1. AI image generators train on stock images ➡ 2. Klarna CEO says "we’ve removed the need for stock imagery from image banks like @gettyimages", instead using AI image generators. Very relevant to the fourth fair use factor (potential market for / value of the original).
10
77
283
16,974
Ryan Groves retweeted
21 May 2024
13 May 2024
her
16
1,230
12,768
566,254
Ryan Groves retweeted
Replying to @DavidDingAI
Being useful for professional creators necessarily involves not exploiting their work. Can you share whether Udio has licensed the music it trains on?
2
3
42
1,141
Ryan Groves retweeted
Update on AI crawlers & @squarespace, having chatted with them: - You are opted-in to allowing AI crawlers by default - When you opt out, they only opt you out of 5 crawlers. You're still opted into crawlers from Amazon, ByteDance (TikTok), Cohere, Facebook, Perplexity etc. - There is no way of manually updating your robots.txt file to block more crawlers The bottom line: none of the estimated ~3 million live Squarespace websites can meaningfully opt out of being scraped for training data by AI companies.
Just found the 'AI crawlers' setting in Squarespace, turned on by default, letting my site's content be used to train AI models. I've been a Squarespace user for years - no idea when this was added. This is the problem with opt-outs.
15
149
305
104,939
Ryan Groves retweeted
20 Apr 2024
This is one of the craziest videos I’ve ever seen. Harry Mack is incredible.
338
2,314
15,285
2,695,050
Ryan Groves retweeted
OpenAI are losing their best and most safety-focused talent. Daniel Kokotajlo of their Governance team quits "due to losing confidence that it would behave responsibly around the time of AGI" Last year he wrote he thought there was a 70% chance of an AI existential catastrophe.
30
72
343
107,325
A good example of copyright laundering out in the wild. Adobe claimed to only train on licensed data, but then added synthetic data generated from the same rivals they were criticizing about copyright. They even compensated users who added generated images. Disappointing.
🚨Scoop from me and @BrodyFord_: Adobe has touted Firefly as a safe and ethical alternative to competitors like Midjourney… but it also quietly trained Firefly on Midjourney images. Adobe’s ‘Ethical’ Firefly AI Was Trained on Midjourney Images bloomberg.com/news/articles/…
1
3
231
Ryan Groves retweeted
this is what AI is for 🔊 sound on

189
891
5,278
811,560
Ryan Groves retweeted
1 Apr 2024
Excited to finally share the news! I’ve been working with Nintendo for the past 3 years to bring Mario Kart to the open road and it’s finally ready to play on the Virtual Boy Pro, coming 2024!

18
6
152
12,760
Well this is likely to be a great conversation between two incredibly thoughtful people.
1
154