Audacity Audiobook Mastering: Full ACX Workflow

The Audacity mastering chain for ACX is four effects in order: Noise Reduction, Compressor, Normalise, Limiter. Apply them in that sequence to meet the ACX audio requirements (RMS, peak, noise floor) and the rest of the spec (format, sample rate, silence). Get one setting wrong and your file fails. Get the order wrong and the measurements shift. This guide gives you the exact settings for each effect, the macro for batch processing, and an honest look at where Audacity's approach has gaps. All settings verified against Audacity 3.7.x (current stable series as of April 2026). These are widely-used community settings, not values prescribed by ACX or Audacity.

The Audacity mastering chain for ACX

Apply these four effects in this order. The order matters because each step changes the measurements that follow. Noise reduction first (before any gain changes), compression second (to control dynamics), normalisation third (to set peak level), limiter last (to catch anything above the ceiling).[6]

Quick-reference settings

Step-by-step detail

Step 1: Noise Reduction.[2] Noise reduction 12 dB, sensitivity 6, frequency smoothing 3 bands.

Step 2: Compressor.[1] Threshold -20 dB, ratio 3:1, attack time 0.2 s, release time 1.0 s.

Step 3: Normalise. Peak amplitude -3.0 dB.

Step 4: Limiter.[3] Type: Hard Limit, input gain 0 dB, limit to -3.5 dB.

Before you start: record a few seconds of silence (room tone) in your recording environment. You will need this noise profile for Step 1. In Audacity, select the silent section, go to Effect > Noise Reduction, and click "Get Noise Profile". Then select your full track and apply the reduction with the settings above.[2]

Why this order? Noise reduction must come before compression and normalisation. If you compress first, you raise the noise floor before you have a chance to reduce it. Compression narrows the dynamic range between your voice and the background noise, making the noise harder to isolate and remove cleanly. Always reduce noise on the unprocessed recording.

Why -3.5 dB on the limiter, not -3.0 dB? ACX requires true peak below -3.0 dBFS.[6] Audacity's limiter works on sample peaks, not true peaks. Setting the limiter 0.5 dB below the ceiling gives you a safety margin for the inter-sample peaks that Audacity cannot detect. More on this gap below.

Why 3:1 compression? A 3:1 ratio is moderate enough to tame dynamic range without squashing the natural variation in narration. Lower ratios (2:1) may not bring quiet passages up enough to hit -23 to -18 dBFS RMS. Higher ratios (5:1 and above) risk making the narration sound flat and fatiguing over long listening sessions. The -20 dB threshold catches most of the speech while leaving quiet breaths and pauses below the compression knee.

The Audacity macro

Repeating four effects manually on every chapter file is slow and error-prone. Audacity's Macros feature lets you save the chain and apply it in batch.[4]

1. Go to Tools > Macro Manager.
2. Click New and name the macro (e.g. "ACX Master").
3. Add each effect in order: Noise Reduction, Compressor, Normalise, Limiter.
4. Set the parameters for each effect as listed in the table above.
5. Save the macro.

To apply it to multiple files: go to Tools > Apply Macro > Files, select your chapter WAVs, and let Audacity process them. Output files appear in a "macro-output" folder.

Limitation: Audacity macros apply effects sequentially but do not verify the results. After batch processing, you still need to measure each file's RMS loudness, peak level, and noise floor yourself. There is no automated pass/fail step.

A note on Noise Reduction in macros: the Noise Reduction effect requires a noise profile captured from a silent section of your recording. When you add Noise Reduction to a macro, Audacity uses whatever noise profile was last captured. This means you need to capture the profile manually before running the macro. If your recording environment changes between sessions, the old profile may not match, and the noise reduction will be less effective or introduce artefacts. For sessions recorded on different days, capture a fresh profile each time before running the macro.

Common errors during macro processing

Files come out too quiet or too loud

The most common cause is running the macro on a file that was already partially processed. If you normalised the file before running the macro, the compressor threshold of -20 dB may not behave as expected because the input levels are different from a raw recording. Always run the macro on your unprocessed, edited files.

Noise reduction creates metallic artefacts

This happens when the noise profile does not match the actual noise in the file. Causes: you captured the profile from a different recording session, your room noise changed (heating turned on, traffic increased), or you used too many dB of reduction. Try lowering the noise reduction to 6 dB and increasing sensitivity to 8. If artefacts persist, the recording environment may be too noisy for post-processing to fix cleanly. See the audiobook noise floor guide for the honest limits.

Macro fails on certain file formats

Audacity macros work best with WAV input. If you feed MP3 or M4A files into the macro, Audacity decodes them first, which adds a generation of quality loss before processing even starts. For best results, keep your edited files as WAV until the final export step.

Audacity version compatibility

The settings in this guide are verified against Audacity 3.7.x (the current stable series as of April 2026). Key notes for other versions:

• Audacity 3.4 to 3.6: The same effect chain works. The Compressor interface changed in 3.5 (the classic compressor was replaced with a new one), but the parameters map directly. If you see different control names, check the Audacity release notes for the mapping.
• Audacity 3.3 and earlier: The Noise Reduction and Limiter effects are the same, but the Compressor used a different interface. The threshold and ratio values still apply.
• Audacity on Linux: All effects are available. LAME MP3 encoding may require installing the lame package separately on some distributions.

If you are unsure which version you have: go to Help > About Audacity. The version number is at the top.

The ACX Check plugin

ACX Check is a free analyser plugin for Audacity that measures your file against the three core ACX specs: RMS loudness, peak level, and noise floor. It does not process your audio. It only measures and reports whether the values pass or fail.

To install: download the acx-check.ny file from the Audacity plugins page, move it to Audacity's Plug-Ins directory, restart Audacity, and activate the plugin via Tools > Add / Remove Plug-ins. Once installed, select your full track, go to Analyse > ACX Check, and read the report.

ACX Check is useful as a quick sanity check after mastering. Two limitations to know:

1. Sample peaks only. ACX Check measures sample peaks, not true peaks. A file that reads -3.2 dB in ACX Check could still exceed -3.0 dBFS true peak and fail ACX's automated analysis. The -3.5 dB limiter setting in this guide compensates for most cases.
2. No noise floor context. ACX Check measures the noise floor of your entire file, including sections where you are speaking. ACX measures noise floor in the silent sections only. If your file has a lot of speech relative to silence, ACX Check may report a different noise floor value than ACX's analyser.

ACX Check tells you whether your measurements are in range. It does not tell you whether your audio will pass human quality review. For the full list of what ACX evaluates, see ACX audio requirements (RMS, peak, noise floor).

Where Audacity falls short

Audacity is a capable editor. For audiobook mastering specifically, it has three gaps worth knowing about.

No true peak detection. Audacity measures sample peaks only.[3] It does not detect inter-sample peaks (true peaks) as defined by broadcast loudness standards. The difference is typically 0.5 to 1 dB. A file showing -3.2 dB sample peak in Audacity could exceed -3.0 dBFS true peak and fail ACX review. The -3.5 dB limiter setting compensates for most cases, but it is a workaround, not a measurement.

No automated verification. After mastering, you need to manually measure RMS, peak, and noise floor for every chapter file. For a 15-chapter audiobook, that is 15 sets of measurements to record and compare against the ACX audio requirements (RMS, peak, noise floor). If you miss one, you find out when ACX rejects it.

MP3 export needs configuration. Audacity includes the LAME encoder on Windows and macOS.[5] MP3 export works, but ACX requires 192 kbps CBR (constant bit rate) specifically. You need to set this manually in the export dialogue: Format Options > Bit Rate Mode > Constant, then select 192 kbps. The default is not 192 CBR.

None of this makes Audacity a bad tool. It works. It requires you to set every value yourself and verify the results manually. For a deep dive into the most common mastering failure, see audiobook noise floor guide. For quick answers about ChapterPass as an alternative, see the FAQ.

ChapterPass as an alternative

ChapterPass runs the entire mastering chain in one step: loudness adjustment, true-peak-aware limiting, noise floor management via speech-aware gating, format conversion to 192 kbps CBR MP3, and silence padding. It processes files locally in your browser without uploading them to a server.

The difference from Audacity is not quality. It is workflow. Audacity gives you four separate effects to configure, a macro to maintain, and manual verification afterwards. ChapterPass applies the same technical corrections automatically and verifies every measurement before producing the output file.

For narrators who want full manual control, Audacity is the right tool. For narrators who want to drop a file in and get a compliant output, ChapterPass does that. First file free. Files never leave your browser. For a broader look at all five options, including Auphonic, Hindenburg, and Source, see the audiobook mastering tools comparison.