The time-to-banger tradeoff

By: Yung Spielburg

Published: 2023-06-08

This is the latest issue of Bit Rate, our member newsletter taking the pulse on music AI. In each issue, writer Yung Spielburg breaks down a timely music AI development into accessible, actionable language that artists, developers, and rights holders can apply to their careers.

Please send feedback at any time by replying to this email, or hopping in #creative-ai in our Discord server.

Last week, we explored user experience (UX) in the music AI landscape, through the lens of Google’s text-to-music model MusicLM.

We’ll continue threading this UX needle today through a slightly different angle, breaking down a key design concept that seems to be permeating the new generation of music AI tools — and perhaps at a cost.

Introducing “time-to-banger”

The more we delve into music software beyond the big five DAWs (Ableton, Logic, Pro Tools, Cubase, and FL Studio), the more we feel the silent hand of “time-to-fun” at work.

In software, “time-to-fun” refers to the amount of time it takes a user to start engaging or playing after downloading an app. Game and mobile developers in particular constantly strive to minimize TTF, competing for every fraction of a second to hook users in.

Many automated music AI tools seem to operate under a similar assumption — namely, that time spent staring at a blank creative canvas, or tediously searching for beats, sounds, or chord progressions, is not “fun.” Instead, what’s more enjoyable is quickly assembling the creative ingredients you need to express your idea, generate a good song starter, and get into a steady creative flow. (In this context, “time-to-fun” might be more aptly named “time-to-banger.”)

The automation x flexibility sweet spot

Many of the new AI tools we’ve been testing attempt to minimize “time-to-banger” by enabling one-click track generation.

In terms of artist workflow, there’s arguably little difference between this automated process and pulling a beat from BeatStars or YouTube. Massive hits of the past, like Desiigner’s “Panda” and Lil Nas X’s “Old Town Road,” have been borne out of this quickfire, scrappy approach to production. AI-powered music creation tools aim to close the skills gap even further, helping users bypass previously necessary training to create music and produce a finished track altogether.

As we covered last week, MusicLM is a step in this direction — though largely a prototype to showcase new underlying raw audio synthesis technology, and not particularly useful in its current form for full-stage music production.

Another example is BandLab’s AI-powered Songstarter tool, where generating initial genre- or lyric-inspired song snippets is literally as simple as rolling a virtual die. Notably though, with Songstarter, you have the option to import a multitrack version of your generated idea into BandLab’s proprietary DAW, BandLab Studio. From there, you can play around with many other production features and presents, and really start to see a full creative picture coming into focus.

This ability to further manipulate the generated material is arguably critical to fostering a long-term creative practice, and suggests that automation and flexibility are not necessarily mutually exclusive when building AI tools.

While efficiency can be a valuable design principle, the way in which a user is brought to “fun” with a creative tool can have an outsized impact on their creative process over time. Too much emphasis on optimizing “time-to-banger” could come at the cost of providing more creative flexibility after the fact, reducing the chance of long-term satisfaction and retention.

Many people in our community and wider network have expressed frustration at the lack of this kind of custom creative functionality in image generators such as Midjourney. Similarly in music, while one-click beat generation is the path of least resistance — leading users straight to the finish line — developers would do well to make sure that that level of automation is not completely collapsing the user’s opportunity for deeper involvement and engagement after the fact.

After all, as any artist will tell you, the creative process can always have unexpected, revelatory pivots or ups-and-downs along the way. As legendary rock engineer Mike Plotnikoff once told a member of our team, “it’s never a straight line.”

Who are we really building for?

Different sets of music AI users also value the “time-to-banger” metric differently.

In our Season 3 report on music AI, we outlined four different user personas for music AI tools, each of which is looking for a different set of outcomes in their experience over time. Some, like solo creators on Twitch or TikTok, might be more intent on automating away music creation altogether, since music sourcing and licensing is just a means to an end to publishing a high volume of content. Others, like professional musicians, are looking to integrate AI into their existing workflows while pushing the boundaries of sound as a whole, and don’t see AI as completely substitutive to their existing work.

Over time, any artist who “catches the bug” of record-making will naturally want to move away from one-click templates, to try and create a sound that is distinctly their own.

This arguably creates a long-term retention challenge for more automated music AI tools, in terms of keeping users interested after arriving at the “time-to-banger” moment.

Some of these tools are attempting to drive long-term engagement and retention by integrating social and discovery elements like charts (e.g. mayk.it). Others, as discussed above, put one-click music AI tools in the context of a larger, DAW-like interface that allows for more granular, extensive creative exploration.

Lowering “time-to-banger” certainly gets more people on the train of creativity, which is a net positive. But without ample room for creators to explore breadth in their creative process, we wonder if these kinds of fully automated AI tools will be able to keep their users around over time. 🤖

Alexander Flores and Cherie Hu contributed editing and fact-checking to this article.

What our members are talking about

Didn’t have time to drop into our Discord server this week? No worries. Stay up to date right here in your inbox with the best creative AI links and resources that our researchers and community members are sharing each week.

Thanks to @colombo, @Kiru, @NatalieCrue, @moises.tech, @DCJ, @aflores, @yung spielburg, and @cheriehu for curating this week’s links.

You can join the community discussion anytime in the #creative-ai channel. (If you’re not already in our Discord server, click here to get access.)

Music and entertainment case studies

AI tools, models, and datasets

Other resources