My Music
The track I worked on this week is called ‘Hold On Tight And Don't Let Go’ and I submitted it to the Naviar Haiku Music Challenge. This is a completely in-the-box production even though the drum set and taiko drum sounds make it sound like I used acoustic instruments. I even played an draft for my wife and she asked, what drums are those? When did you play them? Speaking of her, she is responsible for the album art that accompanies this track! I am very happy with how it turned out. It all started out with playing around with an arpeggiator. It just goes to show how important it is to push through uninspired moments when it seems like a particular sound or technique is not going anywhere. I never really know where I will end up… eventually!
Music and AI
2023 was a year of firsts, including my first trip to a spa. Immediately upon entering I noticed the “spa music.” The aroma and temperature made it truly a multi-sensory experience. Now, I do not subscribe to spa business blogs, but this ad caught my eye when I was searching for articles about music and AI. It is for a music streaming service designed specifically for spas that is customizable for multi-sensory environments and capable of adapting to its surroundings.
I was skeptical of this AI service and decided not to try it after I visited the website and was faced with a click-through screen with all sorts of terms, including one where I had to check a box saying I would not use the results for “commercial purposes” - because I would want to use the sounds in a piece of music and post it. Is that a commercial purpose? I am also turned off, after further reading, to learn the models were trained on “publicly available” data.
A couple of things jumped off the page at me. First, this product went from concept to launch in 45 days. That is breakneck speed even in this new day and age of AI. Second, this product is trained on Rightsify’s own catalog, which makes me assume those license agreements it has with artists had some language in the first place that was vague enough to encompass AI training. Something like, “you give us the right to use your #music to improve our products and services.”
The LimeWire trademark lives on. Speaking of trademarks, I notice the forthcoming product is an “AI-enabled audio workstation, as opposed to a “digital audio workstation.” Is it goodbye DAW, hello AIAW? Regardless of the branding, this development speaks to the trend of using AI to democratize music production. I know I will benefit from this as a musician with little formal training. I already leverage non-AI technology to help me in areas where I have shortcomings.
I am often overwhelmed by sample libraries and packs. When I am faced with a sample search, I will frequently set arbitrary limits for myself, such as only listening to samples with a certain word in the title. I tried this AI platform to see if it could curate a sample pack for me based on a text prompt. The results were not quite on point and some of the sounds were very harsh. But I definitely see the promise and how this technology could improve my search experience.
Yes, we have been living in a world where we knew at least one human was behind the writing of any given song, but it has long been a world where we didn’t know WHICH human(s) wrote it. Dr. Sir Brian May’s point is well-taken, that we are moving to a future where we will have little confidence in the actual capabilities of musicians unless we see the stripped down process of writing and performing live with our own two eyes, and even then we might not believe what we see.
Yet another AI mastering service announcement, but the comment regarding a concern about impersonation of artists caught my attention. In this context, it made me think about the impersonation of other key figures in the #music-making process, namely famous producers and audio engineers, or even production sound mixers, mastering engineers or other instrument-specific technicians who’ve made enough of a name for themselves that someone might try to replicate their style.
SOCAN’s recommendation is to establish a regulatory framework consistent with the underlying goals of Canada’s Copyright Act. This is a thoughtful approach. The U.S. Constitution gives Congress the power to “promote the progress of science” (as in “arts & sciences”) which we now know includes copyrightable subject matter. Consider whether some of the proposed regulatory restraints on #AI technologies would “promote” or hinder “progress” in the field of #music, generally.
I’m surprised to see in this study that the largest category where musicians find #AI useful is songwriting and composition. I know AI tools can be useful for getting a song started, but I would rather stumble on a song idea and use AI to help with more technical tasks. I make music without lyrics, so I hadn’t thought of using AI that way and wonder if it would be useful for that. What doesn’t surprise me is half of respondents are shy to admit whether and to what extent they use AI tools.
Haiku
Dragon fly darts past
With a hurried whirring sound
Six months left to live
Crimson red berries
Bunched up on shedding branches
A father’s lament
Birch tree stump decays
Into the dead grass below
Mushroom emerges
Crumpled boulders
They will never let go of
The soles of my shoes
Crown of golden lace
A well-worn path toward an
Uncertain future
Snow dusted hilltops
Retreat from the flame-tipped clouds
Uncontainable