The Critical Role of Silence and Noise Standards in ACX Audiobook Production

When you produce an audiobook for ACX (Audiobook Creation Exchange), you are entering the largest marketplace for spoken-word content. The platform powers audiobooks for Audible, Amazon, and iTunes, and to maintain a consistent, professional listening experience across hundreds of thousands of titles, ACX enforces strict technical specifications. Among the most important – and often most challenging for new producers – are the rules regarding silence and noise. These standards are not arbitrary; they directly affect whether your manuscript is accepted, how it sounds in a listening test, and ultimately whether listeners stay engaged. In this guide, we expand on ACX’s silence and noise requirements, explain why they matter, and offer actionable steps to ensure your audio files pass inspection every time.

What Are ACX’s Silence Requirements?

ACX does not publish a single "silence rule" but rather a set of expectations that, taken together, define acceptable pause structure. The core principle is that every silence within the audio file must be intentional, clean, and free of unwanted artifacts. Here is what you need to know.

Leading and Trailing Silence

Every file must begin with exactly 0.5 seconds of silence before the narration starts. The same applies to the end: exactly 0.5 seconds of silence after the final spoken word. These "silence headers" and "trailers" are checked by ACX’s automated validation tools. If you submit a file with 0.2 seconds of leading silence or 1.5 seconds of trailing silence, it will be flagged. The exact half-second measurement is critical; you need to trim silence to that precise length using an editor like Audacity, Reaper, or TwistedWave. A common mistake is to assume that cutting the waveform close to the start leaves enough room – but most editors automatically snap to the nearest sample boundary, which can result in slightly off timing. Always use a ruler or selection measurement tool to verify the length.

Inter-Section and Inter-Chapter Pauses

ACX recommends that silences separating chapters or major sections be at least 0.5 seconds in duration. This ensures there is a natural, audible break that signals a transition. Using less than half a second of silence can make the transition feel rushed or sloppy, especially in fiction where chapter breaks carry narrative weight. Conversely, you should avoid inserting more than about two seconds of silence between sections unless the story calls for a dramatic pause. Overly long dead air stands out as unnatural and can cause listeners to check if the audio has stopped. In practice, many producers standardize all chapter gaps to exactly 1.0 seconds for consistency. Whichever duration you choose, stick with it across the entire book.

Silence Within Narration

Inside the narration itself, silence should mimic natural speech patterns. Short pauses between sentences (0.1–0.3 seconds) are normal, as are slightly longer pauses at commas and periods. However, any silence longer than about 0.5 seconds inside a sentence can make the narrator sound hesitant or choppy. If you edit out floor noise, be careful not to compress all pauses into identical lengths, as that sounds robotic. A good rule: listen to the track and focus on whether the silence feels natural. If it draws your attention, it is probably wrong. For example, a narrator reading a dramatic passage might need a 0.8-second pause after a revelation, while a quick back-and-forth dialogue may require only 0.1 seconds between lines. Use editing to preserve these natural variations rather than forcing uniformity.

The Problem of Noisy Silence

Silence that is not truly silent is the most common cause of ACX rejection. Even if your file meets the length requirements, if the so-called silence contains hum, buzz, computer fan, room echo, or breathing, it fails the noise standard. This is where silence and noise standards intersect. Every silence in the audio must have no audible noise. You can test this by turning your volume to maximum and listening to a silent section. If you hear anything other than a gentle hiss of background noise (room tone) that is below -60 dB, you have work to do. Some producers use spectral analysis displays to see noise artifacts – any persistent line in the low frequencies (50–60 Hz) indicates electrical hum that needs to be removed with a notch filter.

Understanding ACX’s Noise Standards

ACX stipulates that the noise floor – the level of background noise when no one is speaking – must be below -60 dBFS (decibels relative to full scale). This is a strict threshold, but it is achievable with proper recording discipline and post-production. Why -60 dB? At that level, room tone is essentially inaudible under normal listening volumes. It provides a clean canvas for the narrator’s voice. The threshold is measured using the RMS (root mean square) level of the silence region, not peak levels. This means that even a brief spike of noise can raise the RMS and cause failure.

Room Tone: Your Baseline

Every room has a unique ambient sound – the hum of a refrigerator, traffic from outside, the gentle hiss of an HVAC system. This is your room tone. Before you record a single page, you should capture 30 to 60 seconds of pure room tone from your recording space. This sample becomes your reference for noise reduction. ACX’s noise standard effectively demands that your room tone, when measured in an empty section of your audio file (like the leading 0.5‑second silence), be at least 60 dB lower than the loudest peaks of your narration. In practice, that means you need to treat your recording environment to reduce ambient noise, then use noise reduction plugins or tools to bring the floor down further. A portable vocal booth or moving blankets can drop floor noise by 10–20 dB in untreated rooms.

Consistent Volume Levels

Closely related to the noise floor is the requirement for consistent RMS (Root Mean Square) loudness. ACX expects an average loudness between -23 dB and -18 dB RMS for the spoken word portions. This ensures that the listener does not have to constantly adjust volume between quiet and loud passages. If your narration varies wildly – for instance, a whisper at -30 dB and a shout at -10 dB – you may need to use compression or volume automation to even out the track. A consistent volume helps the noise floor remain unobtrusive because the ear is not straining to hear quiet words, only to be met with floor noise. Use a compressor with a ratio of 2:1 to 4:1, attack around 10 ms, release around 100 ms, and then apply makeup gain to bring the average into the target range.

Tools and Techniques for Measuring Silence and Noise

To comply with ACX standards, you need to quantify silence length and noise level. Here are the key tools and methods.

Using Audacity’s "ACX Check" Plugin

Audacity has a free plugin called ACX Check (or "ACX Audiobook Checklist") that runs a batch analysis on your file. It checks:

  • Leading and trailing silence length (target: 0.5 seconds each)
  • Noise floor (must be below -60 dB)
  • Peak level (must not exceed -3 dB)
  • RMS loudness range

This plugin is invaluable. Learn more about the ACX Check plugin here. However, note that the plugin measures the noise floor from the first 0.5 seconds of silence, so if your leading silence is not exactly 0.5 seconds, the plugin may give false results. Always ensure the leading silence is correctly trimmed before running the check.

Manual Measurement in Audacity

To check silence length: highlight the silent section and look at the timeline or the selection bar at the bottom (length in hours:minutes:seconds). Trim with the "Truncate Silence" effect or by manually deleting excess. To check noise floor: select a portion of the leading silence, then go to Analyze > Plot Spectrum. The average level in dBFS shows your noise floor. If it is above -60 dB, you need to apply noise reduction. For a more precise measurement, use the "Contrast" tool (Analyze > Contrast) to compare the RMS level of a silence region versus a speech region.

Loudness Metering

For volume consistency, use a loudness meter. Many DAWs have a built-in Loudness Meter effect (e.g., Audacity’s "Loudness Normalization" or third-party meters like YouLean). Measure the RMS of a typical one-minute section. Adjust gain or use a compressor to bring the average into the -23 dB to -18 dB range. Avoid peak normalization because it does not control noise floor and can push silent sections up. Instead, use RMS or LUFS normalization. Many ACX producers recommend normalizing to -20 LUFS (integrated loudness) as a safe target that also satisfies iTunes requirements.

Common Pitfalls and How to Avoid Them

Despite understanding the rules, many producers fail the ACX quality check. Here are the most frequent issues.

Inconsistent Silence Between Chapters

Narrators sometimes leave varying pauses – 0.3 seconds here, 1.0 seconds there. Use a consistent setting in your editing software. In Audacity, the "Truncate Silence" effect can standardize all silences over a certain threshold to a fixed duration. Be careful not to use it on silences inside dialogue, only on inter-sentence or inter-section breaks. A small script or macro can automate this across an entire book project.

Overusing Noise Reduction

Applying noise reduction too aggressively can create "watery" artifacts or cause the narration to sound hollow and digital. The goal is to reduce noise, not eliminate every molecule of room tone. Aim for a natural reduction of 12–18 dB. Use the noise print from your room tone sample, not from a section of narration. Check the result at high volume on headphones. If you hear warbling or background pumping, back off the reduction strength. Some producers apply two passes of light noise reduction (8–10 dB each) rather than one heavy pass to preserve quality.

Ignoring Breath Noise

Natural breaths are part of narration, but loud, gasping breaths become noise. You can lightly gate or fade breaths to reduce their impact. Many producers drop the volume of breath sounds by 10–15 dB. This both cleans up the track and lowers the perceived noise floor during pauses. Just be careful not to remove all breaths – that creates an unnatural "robot" effect. Use a de-breath plugin or manually select breath waveforms and reduce gain. In Audacity, the "Noise Gate" effect with a threshold set just above the breath level can automatically attenuate them.

Failing to Check the Head and Tail

The leading and trailing 0.5 seconds are a common trip point. Your editor might automatically trim to 0.2 seconds if you cut too close to the waveform. Always manually check the start and end. Use the zoom tool to see the waveform at the very beginning; it should be flat (zero amplitude) for exactly 0.5 seconds. Similarly, the end should have a flat tail of exactly half a second after the last syllable decays to silence. Some producers generate the silence using a "Silence Generator" effect rather than relying on trimming, which guarantees the correct length.

Step-by-Step: Preparing Your Final File for ACX

Here is a workflow that ensures both silence and noise standards are met.

  1. Record in a treated space. The better your initial recording, the less post-processing you need. Aim for a noise floor at least -50 dB before processing; that gives you room to clean it further.
  2. Capture room tone. Record 30 seconds of pure silence in your recording environment.
  3. Edit narration. Remove mistakes, replace breaths where needed, and adjust silence lengths between takes.
  4. Apply noise reduction. Use the room tone sample to generate a noise profile, then apply reduction (12–18 dB) to the entire track. Listen critically.
  5. Check leading/trailing silence. Trim to exactly 0.5 seconds at both ends. Use "Silence Generator" if needed. Double-check with selection tool.
  6. Normalize loudness. Use Loudness Normalization (Audacity: Effect > Loudness Normalization) set to -23 dB RMS (or -20 LUFS for integrated loudness). Do not use peak normalization, as that does not control noise floor.
  7. Run ACX Check. Install the plugin and analyze your file. It will flag any failing parameters.
  8. Fix flagged items. If noise floor is too high, re-apply noise reduction or check for hidden artifacts. If silence length is wrong, adjust.
  9. Export to 48 kHz, 16‑bit, mono. ACX requires 44100 or 48000 (48k is recommended), 16-bit, mono WAV. MP3 is not acceptable.
  10. Do a final listening test. Play the first 30 seconds and last 30 seconds at high volume on headphones. Check for any remaining clicks, pops, or uneven noise.

Advanced: Handling Complex Narration Types

Different genres present unique silence and noise challenges. Multi-voice productions need consistent ambient levels across all narrators. Record each actor with the same microphone and room treatment, or apply EQ and noise reduction to match. Matching room tone can be done by capturing a noise print from each actor's space and blending them, but it is easier to have everyone record in the same booth. Audio with music or sound effects (rare for standard ACX but present in some non-fiction) requires that music fades to complete silence before spoken word starts. The same -60 dB noise floor applies to music fadeouts. Use a fade-out that reaches -72 dB or lower to ensure the tail is truly silent. Non-linear editing - if you splice sentences from different takes, ensure the background noise matches. Discrepancies in room tone between spliced segments create audible "clicks" that break the silence standard. Use crossfades of 5–10 ms to smooth transitions.

What Happens If Your File Gets Rejected?

ACX provides detailed feedback on why a file failed. The rejection notice will specify whether the issue is silence length, noise floor, peak level, or sample rate. Do not panic – most issues are fixable. Re-edit the file according to the feedback and resubmit. Because ACX uses automated checking for these technical specs, a second submission often passes quickly if the problem is corrected. However, repeatedly submitting the same error can lead to a strike against your account, so always test thoroughly before uploading.

External Resources for Further Reading

Conclusion: Consistency Is the Key

ACX’s silence and noise standards exist for one reason: to ensure every audiobook listener gets the same high-quality experience, whether they are listening to a thriller from a major publisher or a self-published memoir. Meeting these standards requires attention to detail, a well-treated recording space, and careful post-production. But the effort pays off. A polished file that passes ACX validation on the first try saves you time, money, and frustration. Moreover, it earns you a reputation as a reliable producer. Keep your silences clean, your noise floor low, and your editing consistent. Your listeners – and ACX’s automated checker – will thank you.