The Science of Immersion: Why Audio Storytelling Is So Powerful
Ask someone to describe a book they loved and they'll tell you about pictures: the colour of a room, say, or a face they can still see clearly years later. Then ask where those pictures came from, and watch them pause. The pictures were never in the book. The author wrote sentences, and the reader's mind filled in everything else. Most of what people remember loving wasn't on the page at all. It was in them.
That's the quiet magic of narrative immersion, and audio storytelling happens to be especially good at setting it off. Think of the podcasts people return to every week for years, or the audiobook someone swears got them through a rotten winter. I'd argue they work so well precisely because sound slips past some of the bottlenecks other media put in the way. So this post is a tour of the research behind audio's particular pull. It's also an honest account of why I built EchoQuest the way I did, because a lot of my design decisions trace straight back to these findings.
What Happens in Your Brain During a Story
Brain imaging has turned up something remarkable. When we follow a story, parts of the brain we use for real experience join in. In a 2009 fMRI study at Washington University in St. Louis, Nicole Speer and her colleagues found that as readers moved through a story, different brain regions tracked different parts of it, like where a character was or what they were trying to do. Some of those regions mirror the ones that light up when people carry out similar actions in real life, or merely picture them. Smell works the same way. Another team showed that simply reading words like "cinnamon" or "garlic" activated primary olfactory cortex. And touch gets in on it too: at Emory, people who heard sentences with texture metaphors such as "she had a rough day" showed activity in a region we use to feel texture with our fingers, which the plain "she had a bad day" didn't trigger.
So the brain doesn't merely understand stories. It simulates them. Those simulations are partial and patchy, sure, but they're real, and they produce real emotional responses.
That's what's going on underneath the everyday feeling of being "lost in a book." You aren't really lost. You're running a model. And that model runs on genuine machinery: the parts that handle emotion and plan movement, and sometimes even the circuits that deal with people in your actual life. In a very real sense, stories happen to you, even when nobody's moving and nothing's on a screen.
All narrative works like this. Still, I'm convinced it's strongest when the story reaches you through your ears, because listening leaves the simulation engine with more room to run. Let me explain why I think that.
Reading vs. Listening
Here's a finding that surprises people. When Fatma Deniz and her colleagues at UC Berkeley scanned people listening to stories and then reading the same stories, the maps of meaning across the brain came out almost identical. Your brain understands the words about equally well either way.
So the difference isn't comprehension. It's what else your senses are busy with. When you read, your eyes and visual system are hard at work decoding marks on a page. Skilled readers do this astonishingly fast, of course. Even so, the job of picturing the room has to share space with the job of recognising the shape of the word "room." When you listen, your eyes are off duty. Close them and that whole visual system is free to build the scene the narrator describes. That's my reading of the evidence, and I'll happily stand behind it.
There's some striking data on the emotional side, too. In a study at University College London, 102 people listened to audiobook scenes and watched film and TV versions of the same stories, from A Game of Thrones to The Silence of the Lambs. Afterwards, people said the videos had gripped them more. Their bodies disagreed. Heart rates ran higher while they listened, and so did skin conductance, a classic marker of emotional arousal. The researchers read that as a busier imagination and deeper emotional involvement. In fairness, Audible commissioned that study, so take it with a grain of salt. Even so, it fits everything else we know about simulation.
That's why radio dramas and podcasts build such vivid inner worlds, and why a narrated game can reach a kind of immersion visual games often struggle to match. A visual game gives you the picture and asks you to react. An audio game hands you a seed and lets your imagination grow it. And what grows is yours in a way a rendered image never quite manages to be.
The Role of Voice Performance
How something is spoken changes how it's experienced. Read a sentence slowly, with a small pause before the key word, and you create anticipation that the same sentence on a page simply doesn't have. Vocal performance carries emotional context through rhythm, pitch, pace and silence, and a silent reader has to supply all of that themselves. Take the line "she said his name." Spoken aloud it might land as devastating or as casual. It could drip with sarcasm or shake with fear. The reader has to guess which. The listener is told.
There's evidence that voices carry feelings remarkably well on their own. Across five experiments with 1,772 participants, Yale psychologist Michael Kraus found that people read others' emotions more accurately from voice alone than from voice plus video. Other researchers have since argued that the edge is small. Still, the fact that a voice on its own holds up at all against a voice with a face attached tells you how much feeling it carries.
So a good performance hands you emotional context for free. A reader has to infer that a scene is tense from the prose. A listener simply hears the tension in the voice, without spending any effort to extract it. That spare effort flows back into imagination, and in my view that's a big part of why listeners feel so drawn in.
That's why I spend so much time on narration quality. The ElevenLabs voices on the Storyteller plan aren't just reading text aloud. They're performing it and giving the story room to breathe, something flat text-to-speech can't do. Strip out the performance and much of audio's advantage shrinks. Put it back and you get the full effect.
Getting there was humbling, to be honest. In May I discovered that the premium voices pitched up like chipmunks whenever you sped up playback, so I had to make sure the pitch held steady. Later that month, NPC voices refused to switch at all, because the characters' dialogue wasn't being woven into the narration. Once I fixed that, I matched NPC voices to each character's gender across the whole voice catalogue. And then there was the worst immersion-breaker of all: raw JSON leaking into the spoken narration. Nothing snaps you out of a haunted crypt quite like a narrator reading out curly brackets. That one's fixed too.
Ambient Sound as Cognitive Scaffolding
In EchoQuest, sound does more than accompany the story. It sets the stage before the words arrive. When the low, echoing hum of a cave comes up under the narrator, your brain has already started building the place. The narration then fills a space the ambience has sketched. By the time the GM says "the corridor opens into a vast chamber," your imagination has been warming up for that chamber, and the picture snaps into focus faster and lands harder.
Film composers have used the same trick for a century: music tells you how to feel about what you're about to see. Psychologist Marilyn Boltz showed this neatly in a 2001 study. She paired ambiguous film clips with either upbeat or ominous music, and showed a control group the same clips with no music. The soundtrack bent how viewers interpreted the scene, and even what they remembered of it afterwards, in the direction of the music's mood. Ambient sound works on the same principle. It tells you where you are before you hear what's happening there, and I'd say it's one of the cheapest and most effective immersion tools available. In EchoQuest the beds are synthesised right in your browser, so they cost almost nothing to deliver.
I learned how fragile that scaffolding is, mind you. Back in May I added a forge and a storm to the ambient library, plus an underwater bed, and I taught the ambience to duck politely under sound cues so a sword clash doesn't fight the wind. I also started dropping duplicate cues that fire within 80 milliseconds of each other, because a doubled thunderclap sounds like a glitch, not a storm. Then on May 16 I found the ambient soundtrack was quietly ratcheting itself down to silence after about five turns. A cave that goes dead quiet halfway through a scene feels wrong, even if you can't say why. A couple of weeks later I found it was silent on mobile as well. Both are fixed, and the lesson stuck: the ambience only works if it's always there.
Why This Matters for Blind Players
For sighted players, audio immersion is a choice, a different mode they can opt into. For blind and visually impaired players it's simply how things work. Many bring exactly the skills audio storytelling rewards: close attention to sound and years of building complete worlds from partial information. On top of that, they're used to tracking several sound sources at once and reading emotion in a voice. Lots of blind people have been doing all this every day of their lives.
There's hard evidence for some of it. Researchers in Germany note that people with vision loss can learn to understand speech at up to about 22 syllables per second, while sighted listeners top out around 8. Their brain scans showed parts of the visual cortex pitching in to process that ultra-fast speech. It's also why I think the [ and ] keys, which let you push narration speed up or down mid-scene, matter so much: plenty of experienced listeners want the narrator far faster than any default.
Far from being a compromise, then, I believe audio-first gaming may be the format that plays to blind players' strengths. That's what I designed for. My goal was an RPG where a blind player starts with no disadvantage at all compared to a sighted one, and possibly with an edge. Visualisation skills built over a lifetime of listening transfer directly. So does a sharp ear for tone, and the patience to build a world from sound instead of a glance at a screen.
That goal is why the small timing bugs bother me so much. In May I noticed the choices were being announced while the narrator was still talking, so screen reader users heard two voices at once. Now the choices wait. And narration starts speaking while the GM's reply is still being written, and the next premium clip loads while the current one plays. Dead air is the enemy of immersion, and nobody notices it faster than someone who's listening closely.
Why This Matters for Sighted Players
The flip side holds too. I'd bet sighted players who give audio-first play a real go will find abilities they didn't know they had. Visualisation gets sharper with practice, and your ear for emotion in a voice grows subtler. The mental muscles audio play uses are there in sighted brains as well. They're just underused, because most media aimed at sighted people never asks for them. My hunch is that after a few months of regular listening, audiobooks and radio dramas start to feel richer, because your attention has been retuned. Have you ever noticed a podcast getting more vivid once you closed your eyes? That's the same muscle.
None of this means audio is "better" than visuals. They're different, and each one can deliver a powerful experience. For narrative immersion specifically, though, audio has built-in advantages that the science supports, and I designed EchoQuest to squeeze every drop out of them.
Related reading: Browse all EchoQuest blog posts.
