Hi folks! We're Miriam and Duncan from data sonification studio Loud Numbers, and this is a SPECIAL EDITION of our newsletter that we wanted to send because we felt like we had something to say. You're getting it because you signed up for the regular newsletter, and you can unsubscribe at any time with the link in the footer.

Since we first started getting asked to give presentations and workshops about data sonification, we've always included a 'tools' slide that shows the landscape of sonification tools. But in our latest decks, we've dropped that slide. <insert dramatic LinkedIn pause>. Here's why.

You might have noticed that in our newsletters we always encourage people to send us their work, and they do! We get to see loads of cool sonification projects that people have made (and we often put them in our next newsletter). Recently though, many of the projects we get sent have borne the hallmarks of having been crafted without the use of any specific sonification toolset, but with LLMs instead.

LLMs are powerful, and they'll very happily code you a custom sonification tool that does exactly what you ask it to do – tailored specifically to your project, with no requirement to learn a coding language or a complex software interface. The reality, in 2026, is that learning a specialist sonification tool to make even quite complex sonifications is no longer necessary. This, in a nutshell, is the reason why we've started to cut the tools slide from our presentations and workshops.

But! There are several other edges to the LLM sword. There's the obscene environmental impact. There are manifold ethical and copyright issues. There's the human cost. There's the fact that LLMs are probably the purest form of big tech assholery that we've seen to date, and may crash the global economy at some point soon. In the context of sonification, though, there's one other problem: the work they make is (whisper it) kinda boring.

Boring doesn't necessarily mean bad. On the contrary, LLM-created sonification projects are usually pretty competent. They're somewhat engaging (if you already like sonification, which of course we do). The sound design is pleasant (though often unimaginative, and certainly never intentionally unpleasant). The graphics are clean, and the page is nice to use. They work.

What most LLM sonifications lack, however, is any kind of interesting narrative framing. A reason to exist. An attempt to define what the work means to its creator, and why it's worth spending a few minutes on rather than moving on to the next thing among the billions of other things on the internet. This likely is down to the fact it's so easy to make a sonification project with an LLM. You give it a prompt, it cranks away, it finds bugs, it grinds them to dust, and then voila – one shiny new sonification. Your imagination made real.

That's in stark contrast to the production process without LLMs. Making a sonification project the "old-fashioned" way requires you to spend time learning a specific tool or code library. This is slow, and sometimes boring, and so to propel you through it you need to really want your project to exist, making the hard work worthwhile.

The reasons why people want sonifications to exist are often wildly different on different projects. It could be a story that needs to be told or an emotion that needs to be felt. Like this sonification from Reveal News back in 2016 memorialising Florida's Pulse nightclub shooting victims. Or Edward Martin's vandalising of Sigur Ròs's Hoppípolla to sonify shocking declines in global wildlife populations.

The reason doesn't need to be heavy, though. It could be playful – like when we made a generative music engine for Andy Kirk's mammoth Seinfeld project last year, because we were excited to work with the incredible dataset that Andy had collated and wanted to make some music out of it.

In fact, a reason can be circular: it can simply be the desire to create something really good that matters because lots of hard work went into it, by someone who cared about how it turned out. Designer Giorgia Lupi's recent New York Times essay on the value of graft in creative work and what gets lost when we skip the slow part is well worth a read here.

When it's easy to make a sonification – when you don't need to learn a tool – you end up in a situation where people make sonifications because they can, not because they have a compelling must which drives them through the sometimes-tedious tool-learning process.

This isn't about gatekeeping – it's good that more sonifications are being made. More sonifications in the world means more listeners exposed to sonification. More creators experimenting with sonification as a medium is good, too, because they might come up with cool, new, weird ideas (especially if they're not following the same tool-dictated pathways that others are).

But if you want to create a project that stays with you, one that you'll still be proud of in five years' time, then here's what we recommend. First, don't start by throwing a prompt at an LLM. Instead, start by figuring out what you want to say – what your story is, your narrative framing. Then think about who needs to hear it, and how you can share that story in a way that will resonate with them. Finally, put some thought into what you want your audience to do once they've heard it. Write those all up in a doc, in your own words, and only then take it to your favourite sonification tool – even if that tool is an LLM.

Once your tool of choice has made a first version, don't stop there. Challenge it. Try different data mappings, especially those that feel intuitive, or (even better) those that feel a little odd or dangerous. Focus on the details: the sound design, the visual design, even the typography. Don't let an LLM write any text unless you want the text to sound like it was written by an LLM – write it in your own voice instead. Once it feels right, once it makes you feel something, test it on friends and family – does it feel right to them? Does it make them feel things? Does it feel to them like it was made by you, as opposed to a tool?

There's much, much more to it than that, but at root, what we're describing here is basically a design process. Skip it, and you'll get empty-feeling work that's exciting for a few minutes and is then discarded in favour of the next new shiny thing. Without a story that you want to share, a sonification project ends up being more about novelty than narrative. And novelty rarely stays with you, especially when there's more content than ever fighting for your poor, fractured attention span.

We've got a workshop coming up where we'll be talking about the storytelling process in a lot more detail – it's called Storytelling With Sonification and it'll be held across two sessions on Zoom, on 30 Nov and 4 Dec 2026 (click through for timings). It's aimed at people who already have some familiarity with sonification, and want to deepen their practice. We have a full-price rate, and a discounted rate for students and low-income folks. If you want to find out more, then head over to loudnumbers.net/apply.

We'd love to hear your thoughts on the above - is it something you're seeing too? Do you agree with our diagnosis? What does it mean for the field in the longer term, and for your work specifically? Get in touch - either at [email protected], or in the Decibels data sonification community.

Otherwise, that's all from us today. We'll be back in touch with a regular newsletter again soon. Thanks for your time!

- Duncan and Miriam