Music By AI
Music is everywhere around us, in our media, at our events, in our heads and in our history. With new advancements in technology we can see a popular uprising in the use of AI assisted music production. From cartoon character song covers to Drake’s “Heart on my Sleeve”, AI assisted music has taken the internet by storm. Join us as we explore AI’s implementation of Music.
Tiktok AI Covers
If you have been active in the slightest on Tiktok, surely you have come across one either an AI cover or an AI generated song. One popular example of an AI generated song is “Heart on my Sleeve” by Ghostwriter. The song uses the voices of Drake and The Weekend despite the two having nothing to do with the production. Going viral on Tiktok around April of this year with many fans reposting and enjoying the lyrics, the song was submitted for best rap song and song of the year for the Grammys. These prizes are awarded to the writer of the song rather than the performer, which means Ghostwriter was completely eligible for the awards despite the computer generated vocal performance. Arguments were made highlighting the AI cover’s eligibility to be submitted for the Grammys due to the fact it was not performed by a person, however as the lyrics were written by a human it still remained legible.
Another recent cover spreading across Tiktok is Lana Del Rey’s cover of Mistki’s “My Love Mine All Mine”. I played this song for a friend of mine saying Lana had done a cover of Mitski’s latest song and he said it sounds really great and natural to Lana’s style, then only did I reveal that it was an AI generated cover. Personally seeing more of the AI cover rather than the original song circulating Tiktok, I feel a sense of despair for the original and for Mitski. It was her hard work and effort to write and perform the song yet a robotic generation of the product is what seemed to have made it to the media.
Artists Working with AI
In a different form of AI music usage, some artists have opted to use the technology in their official music production. Canadian musician and singer-songwriter Grimes was one of the first to commercialise the usage of AI music software. Grimes introduced a program called GrimesAI allowing artists to use her voice samples in their creations. If Grimes approves of the creation from the program, she can legally produce it and is subject to a 50% royalty split with the creator. The software used in the program is Elf.Tech which is a generative AI operating system that has been trained with Grimes’s voice. Some of the songs made by using GrimesAI have already been released on streaming platforms. An example of one of these songs is “Cold Touch” by KITO and GrimesAI.
Another artist using AI assistance in their music is Holly Herndon. Similar to GrimesAI, Holly introduces an “AI vocal twin”, Holly+. With Holly+ creators are able to transform their own voices into Holly’s. The platform also allows users to create songs in different languages, this is all thanks to a technique Holly calls “spawning”. Similar to the more known technique of sampling in music production, spawning uses a similar method. Spawning is what allows Holly+ to create a wide range of vocals that were not sung before, this is done by sampling and training from a set of recordings that were previously sung by Holly.
While initially seeming quite dystopian to be lending your voice for public usage and manipulation, Holly’s intent is quite opposing to that. Holly believes that this technology can inspire fellow artists to create in a different fashion amidst the technological revolution changing the way we process art.
With new outlooks of this technology like GrimesAI and Holly+, almost anyone who is capable of writing lyrics can create music. With a little creative application any individual can express themselves through song without having to sing a single verse. This also introduces a new output for current producers. In a sense we are redefining what it means to be a musical artist.
Fully AI Generated Music
The examples previously discussed have all involved a large form of human input, using AI more as a tool. How about a fully AI generated song? From lyrics, to tempo, rhythm, chord progression and singing, an AI who goes by the name Anna Indiana wrote and performed “Betrayed by this town”. The results? A complete disaster. The melody of the song is unnaturally robotic and flows awkwardly without any predictable repetition. The lyrics themselves are quite shallow and lack artistic flare. The awkward emphasis of different parts of words also contributes to the inhumane feeling.
A slightly different example of music generated by an AI is A song called “Daddy’s Car” from Sony Computer Science Research Lab. The song was fully composed by an AI called FlowMachines back in 2016. Similar to the procedure discussed earlier, FlowMachines analyses a large database of existing songs to understand and learn the patterns and structures of them. In this case FlowMachines was trained on a dataset of songs from The Beatles. The final composition was performed by professional musicians. Compared to the previous song, “Daddy’s Car” had a much better outcome. The melody feels much more natural and musical. It feels almost haunting how the AI was able to capture the essence of The Beatles. However, I do believe that the human performance of the composition is what allowed the piece to feel a little more natural.
A Personal Note
Whether we agree with this technology or not, I think that if the artist believes their voices are being used for good to help others express themselves, it may be a new chapter for music production. In light of fears that AI generated music will replace our artists today, I believe there is a far way to go before it can fully create quality pieces. Music is a form of human expression, it requires feelings to be communicated through quality performances. Along with this, live musical performances are one of the best ways musicians connect with their audience. Listening and experiencing musical expression in person from another human is something AI can not take away from us. It is exciting to see where music by AI will take us, but we must be weary of how far we go, as to not lose the artistic human touch and connection that music brings us.
[Written By: Nurrania binti Alfian, Edited By: Teoh Jin]
