Sony CSL (Paris)
7 HN mentions · mentioned in 7 months, 2016–2025
3:01 · 3.3M views · Sep 19, 2016
Sony CSL's Daddy's Car is a pop song composed with AI in the style of the Beatles, with lead sheets generated by software from Beatles-style training material. Humans produced, arranged and performed it.
Cited as an early example of AI-composed music, usually with the caveat that humans did much of the production.
7 comments from Hacker News
My take (as someone working in the area) on how/why we are encouraged by this progress: The model is unconditional i.e. any concept of form at all was learned directly from data, with no real structural hints to what it should learn. Generally you get much stronger "global structure" the more prior information you provide either in the model itself or in the input/targets (such as chord constraints, etc.). This is one reason that harmonization tasks often sound notably better than pure generation - the "backbone" in harmonization was human provided, whereas in pure generation the model needs to be self-consistent. The analogous task in language would be character level language modeling, whereas something like harmonization or conditional generation from chords would be more akin to machine translation. Character level language models may be a bit wonky looking when generating, but any structure at all was purely model discovered from a simple rule (generally, maximum likelihood), whereas in conditional models a lot of extra help is given by the conditioning variable. This model is a pretty big jump in quality from previous efforts in the same vein of unconditional generation, and should combine well with techniques for conditioning in sequence models from all over the research community (translation, speech recognition, question answering, summarization, speech synthesis). On top of that, the model is IMO quite elegant, and much simpler to understand than many other attempts
OP and others may be interested in this approach by Sony CSL from ~2012 using constraint Markov chains for matching lyrical style with explicit rhyme constraints, including (e.g.) Bob Dylan style with constraint of Beatles' Yesterday [0]. This is the precursor to the work that brought on "Daddy's Car"[1], though the techniques appear to have changed a bit reading their recent publications. I find their paper on recent approaches for music synthesis with style matching or other constraints [2] quite readable, and really good! I am hoping to build out a version of what is described by the paper. There is also a very nice talk by F. Pachet on this subject here [3], and his older videos are good too (they should be in the recommendations of YouTube). Still reading about DeepRhyme, but it seems great. Also a fantastic writeup! [0] [LINK] [1] this video [2] [LINK] [3] [LINK]
Here's a song by Sony researchers that was composed by AI, and then a human wrote the lyrics and produced it. The intent was to create a Beatles sound, and I think they did quite well. this video Add some Grover-like text for the lyrics ( [LINK] ), and you're most of the way there. Then use a deepfake singer ( [LINK] ), and you're most of the way to done. Depending on the audience, particularly their level of attention or focus on the music, you can probably get away with playing a completely un-human pop song that people will like. [LINK] I find the possibilities here very interesting, but I also greatly fear that this will become the norm the same way that nearly every modern song uses autotune.
I mean we're not far off with Vocaloid. Ever seen a Hatsune Miku concert [0]? All that needs to be met is "generate lyrics and song by A.I" and I've already heard a number of A.I generated that could work well enough as backgrounds for a vocal-heavy track [1]. The hardest part would be training a neural network to generate Miku's vocals and maybe creating the lyrics for the song. [0] [LINK] [1] Just one example: this video
Humans are fallible in a way computers are not. One could argue any creative process is an exercise in fallibility. More interestingly, humans are capable of assessing the results of their "neural misfires" ("hmm, there's something to this"), whereas even if we could make a computer do such mistakes, it wouldn't know its Penny Lane from its Daddy's Car[0], even if it managed to come up with one. [0] this video