Authors ask this question in roughly one shape: should I put music behind my audiobook? The word doing the damage is behind. It assumes music is a layer you switch on and leave running, and once that is the mental model, both possible answers are wrong — one gives you a book with a loop under it, the other gives you a book with nothing at the seams.
Productions that use music well are not running it in the background. They are placing it, a few seconds at a time, at moments chosen in advance.
No audiobook needs background music, and the productions that use music well are not using it as background. Music earns its place at a change — a chapter boundary, a scene shift, an emotional turn, an ending — and it should be gone before the next line needs your full attention. If a cue is not marking a change, it is furniture, and listeners stop hearing furniture.
What listeners are actually objecting to
Search this question and you will find the split reported as a matter of taste: some listeners like music, some do not. That framing hides the real pattern. The two camps are describing different products.
A bed is continuous. It starts near the top of a chapter and runs underneath the narration more or less until the chapter ends. It is cheap to produce, and it fails for a reason that has nothing to do with the music being bad: a signal that never changes stops carrying information. Attention files it as room tone, and from there it is either ignorable or irritating — and on a phone speaker in a car, it is the thing making words harder to catch.
A score is discontinuous. It appears at a boundary, does one job, and gets out. Because it changes, it stays informative. A listener who could not tell you there was music in a production they enjoyed was almost certainly listening to a score, and that is the compliment the craft is aiming for.
When someone says they hate music in audiobooks, they nearly always mean beds. When someone says a production felt cinematic, they nearly always mean a score. The disagreement is about production method, not preference.
Ready to try it on your own book?
Start your first chapter free →The four moments a score earns its place
Chapter open and close. The most useful cue in audiobooks, and the least controversial — it is also the only placement the retail specs explicitly bless. A few seconds of music at a chapter boundary tells a listener the unit has changed, which matters more in audio than in print because there is no page turn and no white space. It also gives someone who is resuming after two days a moment to reorient before the prose starts.
A scene or location change. Prose marks these with a line break. Audio has no line break, so an unmarked scene change lands as confusion — the listener spends the next two sentences working out that the characters, the place, or the year has moved. A short cue does what the white space did, and it does it faster than a pause.
An emotional turn the prose has already made. Music can raise the temperature on a turn the text earned. It cannot manufacture one. This is the placement that separates scoring from decoration: the cue arrives after the sentence that did the work, confirming it, rather than arriving early and telling the listener what to feel about a line they have not heard yet. Score the aftermath, not the setup.
The ending. The last minutes of a book are the one place where a longer, more present cue is welcome, because there is no more information coming that the listener needs to protect. Endings are where audiobooks are allowed to behave like film.
Where music reliably hurts
Under dense dialogue. Dialogue is where the listener is doing the most work — tracking who is speaking, holding two intentions at once. Adding a second continuous stream to that is the fastest way to make a scene tiring. In a full-cast production, the voices are already carrying the texture that music would otherwise supply.
Under exposition the listener has to remember. A rule in a fantasy magic system, the layout of a house in a mystery, a date that pays off in chapter twenty. If a listener has to hold it, give them a clean channel to hold it in.
Anywhere it runs long enough to stop being noticed. This is the bed failure stated as a rule. Any cue that outlives the change it was marking has become background, and background is the thing you were trying not to make.
Three approaches, compared
| Dry narration | Music bed | Scored production | |
|---|---|---|---|
| Where music appears | Nowhere, or credits only | Continuously under narration | At boundaries and turns only |
| What it costs to make | Nothing extra | Low — one loop, laid in | Higher — every cue placed against the text |
| Effect on comprehension | Neutral | Negative on phone speakers and in cars | Neutral to positive; cues mark structure |
| How listeners describe it | "Clean", "just the book" | "Distracting", "why is there music" | "Cinematic", often without noticing the music |
| Genre fit | All; the default for nonfiction | Weak everywhere | Fiction, especially gothic, fantasy, thriller |
| Risk | Scene changes land flat | Reliably the complaint people post about | Cues placed on the wrong beat read as manipulative |
The middle column is the one worth avoiding. It is the cheapest way to add music and the only one with a systematic downside.
Genre moves the line
The test does not change by genre, but the tolerance does.
Gothic, horror, and fantasy absorb the most score. These are genres where atmosphere is part of the contract — a reader picking up gothic fiction has already agreed to be put in a mood — and where locations are doing narrative work that a cue can help establish. Our production of Frankenstein leans on this: the Arctic framing and the laboratory are different sonic places, and the score is part of how you know which one you are in.
Thriller and mystery use music mostly structurally — chapter boundaries, the turn at the end of a scene — because these genres run on withheld information, and a cue that telegraphs a reveal costs more than it adds.
Literary fiction rewards restraint hardest. Wuthering Heights is the useful case: the score is sparse and low, and most of the production decisions are about where music does not go. Restraint is a scoring decision too, and it is the harder one.
Memoir, business, and self-help are the genres where the honest answer is usually none. The listener came for information in a voice they trust. Credits music at the open and close is the whole appropriate budget.
What the retail specs actually allow
This is the part most discussions of the question skip, and it settles more of it than any opinion.
ACX's audio submission requirements state that music may be added to opening and closing credits, and that it must not obscure the narration. That is the retail floor written down: credits music is expected, anything under the narration is your risk to manage, and the standard it is judged against is intelligibility.
The same page sets the technical envelope every audiobook has to land in — RMS volume between -23 dB and -18 dB, peaks below -3 dB, a noise floor under -60 dB, delivered as 192 kbps CBR 44.1 kHz MP3. Those numbers apply to the finished file, which means a music cue and the voice are spending from the same budget. It is a useful reality check on any plan that involves music running constantly.
Worth stating plainly, because it trips authors up: ACX requires human narration and does not accept unauthorized AI or text-to-speech, so it is not a distribution path for AI-narrated work regardless of how it is scored. Its spec sheet is still the most widely adopted description of what a retail audiobook file looks like, which is why it is worth reading even when you are publishing elsewhere. If you are working out where AI-narrated productions can actually go, we cover that in where to distribute an AI audiobook.
Levels are a different question
Everything above is about placement — which moments get a cue. There is a second question underneath it, which is how loud that cue sits once it is there, and the two get argued as one all the time.
They are separable. A perfectly placed cue mixed too hot still makes a word harder to catch, and a badly placed cue mixed politely is still clutter. We wrote the level half separately in how loud audiobook sound effects should be — the short version is dialogue-first, always. For how music sits alongside effects and voices in a finished production, see our overview of sound design in audiobooks.
Where this gets decided in practice
On Midsummerr, the scoring question is a tier choice rather than a post-production task. Full Cast ($3.75 per 1,000 words, self-serve) gives every character a distinct voice and no music or effects. Full Production ($5 per 1,000 words) adds the music, the sound effects, and a cinematic mix for every chapter. The gap between them is exactly the decision this article is about, and it is $1.25 per 1,000 words.
Cues are placed against the text rather than laid under it, and every one of them is editable — you can move a cue off a line it was crowding, or take it out entirely, without re-rendering the chapter. That matters more than it sounds, because scoring is a judgement call that gets better on the second pass. See pricing for the full tier breakdown.
Hear it
This argument does not survive being described, so the useful move is to test it against your own ear rather than take the reasoning on trust.
Play a chapter of Frankenstein and try to notice every place music enters and leaves. Most people find that harder than they expect, which is the point — cues that mark a change are doing their job when they read as structure rather than as music. Then play Wuthering Heights, where there is far less of it, and see whether the restraint reads as absence or as taste.
Both are complete productions in our public listening library, not demo reels.
FAQ
Do audiobooks need background music?
No. No audiobook requires music, and the majority of retail audiobooks have none beyond credits. The question worth asking is not whether to add music but whether specific moments — chapter boundaries, scene changes, the ending — would be clearer with a cue marking them.
Does background music hurt audiobook sales or reviews?
There is no reliable public data isolating music as a variable, and any figure claiming a precise revenue effect should be treated with suspicion. What is observable is that listener complaints cluster on continuous beds and on levels, not on the existence of a score. Scored productions with cues at boundaries draw the "cinematic" description; beds draw the complaints.
Is music allowed in an audiobook on retail platforms?
Yes, with conditions. ACX permits music in opening and closing credits and requires that it not obscure the narration. Music elsewhere in the file is not prohibited outright, but it is held to the same intelligibility standard and has to fit inside the same loudness envelope as the voice.
How much music should an audiobook have?
Less than authors expect. A useful starting budget for fiction is a cue at each chapter open and close, one at each significant scene or location change, and a longer one at the ending. If a chapter has music running for more than a small fraction of its length, it has probably drifted into being a bed.
What is the difference between a music bed and a score?
A bed runs continuously underneath narration. A score appears at chosen moments and stops. The bed is cheaper to produce and is the version listeners object to; the score costs more attention to place and is the version that reads as production value.
The short version
Music in an audiobook is a placement decision, not a switch. Put cues where something changes — chapter boundaries, scene shifts, the turn after the prose has done its work, the ending — and take them out everywhere else. Continuous beds are the one approach with a consistent downside, and they are the reason the question has a bad reputation.
Hear a scored production, then play something with almost none of it, and the decision stops being theoretical.
