Audiobook Narration: Meeting ACX Standards
# Audiobook Narration and ACX Standards: What You Need to Know Before You Record
Getting audiobook audio accepted by ACX (Audiobook Creation Exchange), the production and distribution platform behind Audible and Amazon, comes down to meeting a specific set of technical requirements. Those requirements exist for good reason: listeners expect consistent volume, clean audio, and professional sound whether they're streaming on a phone, playing through a car stereo, or using earbuds on the train. If your files don't meet spec, they get rejected, and you're back to re-recording or re-processing.
The ACX technical standards are not especially difficult to hit, but they do require the right recording environment, proper gain staging, and a clear understanding of what each measurement actually means. We work with narrators, authors, and publishers on [audiobook production](https://animusstudios.au/services/audiobook-production) regularly, and the same problems come up again and again: rooms that are too noisy, levels that are too hot or too quiet, and files that haven't been properly edited or measured before submission.
This guide covers every major ACX requirement, explains what each one means in practical terms, and tells you how to hit the targets consistently.
---
The ACX Technical Requirements at a Glance
Before getting into the detail, here are the core numbers you need to know:
- RMS (average loudness): between -23 dBFS and -18 dBFS per finished file
- Peak level: no higher than -3 dBFS
- Noise floor: -60 dBFS or quieter (measured in silence, no signal)
- File format: MP3 at 192 kbps or higher, or WAV/AIFF at 44.1 kHz, 16-bit or 24-bit
- Room tone: each file must begin and end with 0.5 to 1 second of room tone (not silence, not digital black)
These are the hard limits. Miss any one of them and ACX will reject the submission. Meet all of them with clean, well-performed audio and you're in good shape.
---
What RMS Actually Means and How to Hit It
RMS stands for Root Mean Square, which is a way of measuring the average energy of an audio signal over time. For ACX, the target is between -23 dBFS and -18 dBFS. Think of it as the perceived loudness of your narration across the whole file, not just the loudest moments.
The most common mistake narrators make is recording too quietly and then boosting the level in post, which raises the noise floor along with the signal. The better approach is to set your gain correctly at the source so that your natural speaking voice lands consistently in the right range. In practice, that means your recording level should be sitting somewhere around -18 to -12 dBFS on the meter while you're speaking, leaving headroom for peaks without clipping.
To measure RMS accurately, use a loudness meter that reads in dBFS RMS. iZotope RX, Adobe Audition, and Reaper all have built-in tools for this. Waves WLM Plus is a dedicated loudness meter that works well for this purpose. Run the measurement across the entire finished file, not just a section, and check that it falls within the -23 to -18 dBFS window. If it's too quiet, a gentle gain increase or light compression can bring it up. If it's too loud, you may need to reduce your input gain and re-record, or apply a small amount of gain reduction in post.
---
Peak Levels and Why -3 dBFS Is the Ceiling
The peak level requirement, no higher than -3 dBFS, is about preventing clipping and distortion on playback. Digital clipping happens when a signal exceeds 0 dBFS, and even peaks that get close to 0 dBFS can cause inter-sample peaks that distort on certain playback systems and after MP3 encoding.
Keeping peaks at or below -3 dBFS gives you a safety margin. A true peak limiter set to -3 dBFS, applied as the last step in your processing chain, is the cleanest way to ensure compliance. FabFilter Pro-L 2 and the Waves L2 are both reliable options for this. Set the ceiling to -3.0 dBFS, enable true peak limiting if the option is available, and let it catch any stray transients without audibly squashing the narration.
It's worth noting that the -3 dBFS ceiling and the -18 to -23 dBFS RMS target are not in conflict. A well-recorded narrator speaking at a consistent level will naturally have peaks that sit well above the RMS average, and that dynamic range is normal and expected. You're not trying to compress everything flat.
---
The Noise Floor: -60 dBFS and Why It's Harder Than It Sounds
The noise floor requirement is where most home narrators run into trouble. ACX requires that the noise floor of your recording, measured during a section of silence with no signal, sits at -60 dBFS or quieter. That is a very quiet room.
To put it in perspective, a typical untreated bedroom with air conditioning running, traffic outside, and a computer fan nearby might measure anywhere from -40 to -50 dBFS. That's 10 to 20 dB louder than the ACX limit, and it will result in rejection.
What Contributes to a High Noise Floor
- HVAC and air conditioning: The single biggest offender in most homes and offices. Turn it off during recording if possible.
- Computer fans: Position your computer outside the recording space or use a long cable to keep it away from the microphone.
- Electrical hum: Ground loops and cheap preamps can introduce 50 Hz hum. Use quality interfaces like the Universal Audio Apollo or Focusrite Scarlett series.
- Room reflections and reverb: Not technically noise, but an untreated room will add colouration that makes the recording sound unprofessional.
- Microphone self-noise: Condenser microphones like the Neumann U 87 or AKG C414 have very low self-noise (around 12 dBA), which helps. Budget condensers can have self-noise of 20 dBA or higher, which contributes to the noise floor.
Measuring Your Noise Floor
Record 10 seconds of silence in your recording space, with your microphone active and your gain set as it would be during narration. Import that file into iZotope RX or your DAW and measure the RMS level of that silent section. That number is your noise floor. If it's above -60 dBFS, you have a problem to solve before you start recording.
Noise reduction plugins like iZotope RX's Voice De-noise or Spectral De-noise can help reduce residual noise, but they are not a substitute for a quiet room. Over-applied noise reduction introduces artefacts that sound unnatural and can themselves cause rejection.
---
Room Tone: The Requirement Narrators Most Often Miss
Every finished audio file submitted to ACX must begin and end with 0.5 to 1 second of room tone. Room tone is the natural ambient sound of your recording environment with no signal, not digital silence (which is absolute zero, completely flat), but the actual acoustic character of the space you recorded in.
The reason for this requirement is practical: it gives the listener's playback system a moment to adjust, and it allows for clean editing and assembly on the production side. It also makes it easier to detect whether a narrator has artificially silenced sections of their recording, which can cause audible clicks or unnatural transitions.
The easiest way to handle room tone is to record a full minute of it at the start of each session, before you begin narrating. Then, when editing, trim each file so that it starts and ends with a short section of that room tone rather than a hard cut to silence.
---
Microphone and Room Setup for ACX-Compliant Recording
The microphone you choose matters, but the room matters more. A Neumann U 87 in a bad room will still fail the noise floor test. A Shure SM7B in a well-treated space will pass.
For narration, large-diaphragm condenser microphones are the standard choice because they capture the full frequency range of the voice with detail and presence. The Neumann U 87, AKG C414, and Audio-Technica AT4050 are all solid options. The Shure SM7B is a dynamic microphone that works exceptionally well for narration in less-than-ideal rooms because its tighter pickup pattern and lower sensitivity make it less susceptible to room noise and reflections.
Position the microphone roughly 15 to 30 centimetres from your mouth, slightly off-axis if you're prone to plosives, and use a pop filter. Speak at a consistent distance throughout the session. Inconsistent mic distance is one of the main causes of level variation that makes RMS compliance difficult to achieve.
If you're recording at home and your space is not treated, a portable vocal booth, heavy curtains, and recording inside a wardrobe full of clothes are all legitimate options that can bring your noise floor down enough to meet the -60 dBFS requirement.
---
Editing, Mastering, and Final Delivery
Once narration is recorded, the editing process for audiobooks is detailed work. Every breath, mouth noise, stumble, and long pause needs to be addressed. ACX does not specify a maximum for breath sounds, but excessive breathing is flagged by quality reviewers and can result in rejection on quality grounds even if the technical specs are met.
After editing, the processing chain for ACX delivery is straightforward: a gentle high-pass filter to remove low-frequency rumble (roll off below 80 Hz), light compression to even out level inconsistencies, a de-esser if sibilance is an issue, and a true peak limiter set to -3 dBFS at the end of the chain. Measure the final RMS of each file before export and adjust gain as needed to land in the -23 to -18 dBFS window.
Export at 44.1 kHz, 24-bit WAV for archival purposes, then encode to MP3 at 192 kbps for ACX submission. Keep your WAV masters. If ACX updates their requirements or a publisher needs a different format, you'll want the full-resolution files.
Our [vocal recording](https://animusstudios.au/services/vocal-recording) and [audiobook production](https://animusstudios.au/services/audiobook-production) services are set up specifically to meet these standards, with treated rooms, professional microphone options, and a mastering process built around ACX compliance.
---
Meeting ACX standards is not about chasing numbers for their own sake. It's about delivering audio that sounds professional, plays back consistently across every device, and gives your listeners the experience they're paying for. Get the room right, set your levels correctly at the source, edit thoroughly, and measure everything before you submit. Those four steps will get you through ACX quality control and onto Audible with a file you can be proud of.