Audio Format Guide for Audiobooks

ACX mandates the following for all audiobook submissions: MP3 format (no WAV, FLAC, or AAC accepted), 192 kbps CBR (Constant Bitrate — VBR files are rejected), mono (not stereo), 44.1 kHz sample rate, RMS loudness between -23 dBFS and -18 dBFS, peak level no higher than -3 dBFS, noise floor below -60 dBFS. Every chapter file must meet all of these specifications.

Step 1: Record in WAV, 44.1 kHz, 24-bit, mono. Using 24-bit during recording gives you headroom — you can bring the levels up during mastering without noise artifacts. Step 2: Edit in Audacity, Adobe Audition, or your DAW — remove breaths, room noise, and mistakes. Step 3: Apply noise reduction to achieve the -60 dBFS noise floor. Step 4: Normalize to -23 LUFS RMS and -3 dBFS peak. Audacity's ACX Check plugin can verify compliance.

After mastering your WAV files, convert to MP3 using the WAV to MP3 converter at audioutils.com. Select 192 kbps CBR in the encoding options. The free tier converts files up to 10 MB; Pro handles up to 500 MB (sufficient for chapter-length audio at WAV quality). The output MP3 will be 192 kbps CBR mono — the exact format ACX requires. Verify the output file meets specs before uploading the full project.

Findaway Voices (Spotify for Audiobooks, library platforms) accepts WAV or MP3, requires -23 LUFS average loudness, -3 dBFS peak, and -60 dBFS noise floor — the same loudness standards as ACX. Authors Republic accepts MP3 at minimum 192 kbps or WAV. Kobo Writing Life accepts MP3 at 192 kbps minimum. In practice, mastering to ACX spec and converting to MP3 192 kbps CBR works for all major platforms.

Noise floor too high: record in a quieter space, apply noise reduction plugin. Loudness too low: normalize to a higher RMS level. Loudness too high: apply a limiter to reduce peaks, then normalize down. Wrong format: convert WAV to MP3 192 kbps CBR. Stereo file submitted: re-record or convert to mono by mixing both channels. Chapter too short: ACX requires a minimum of 15 seconds of audio per file.

ACX's format rules are not arbitrary — each serves a distribution reason worth understanding so you do not fight them. MP3 rather than WAV or FLAC: Audible needs a compact, universally-decodable file that streams to phones and dedicated players; 192 kbps MP3 is transparent for spoken word (voice compresses far more easily than music) while staying small. CBR rather than VBR: constant bitrate guarantees accurate seeking and duration display across every Audible player, and ACX's ingestion pipeline rejects VBR outright because variable files can misreport length. Mono rather than stereo: narration is a single voice with no meaningful stereo information, so mono halves the file size for identical audio and avoids channel-balance problems on phone speakers and one earbud. 44.1 kHz: the standard spoken-word sample rate, more than enough for the voice's frequency range. Master to these targets from the start rather than converting a stereo music-style file at the end.

ACX has structural requirements beyond the audio specs, and missing them causes rejection just as surely as a bad noise floor. Room tone: every file must open with 0.5 to 1 second and close with 1 to 5 seconds of silent 'room tone' — the recorded ambient silence of your space, not digital zero — so Audible's player has natural padding between sections. Opening credits: the first file must announce the title, author, and narrator ('...written by..., narrated by...'). Closing credits: the final file must include a closing announcement. Retail audio sample: you provide a 1-5 minute representative sample (often from the middle of a chapter) that Audible uses as the preview. One chapter or section per file: do not combine chapters, and name files clearly in order. Each file must be at least 15 seconds and no longer than 120 minutes. Getting the structure right the first time avoids re-submission delays.

Reaching -23 to -18 dBFS RMS, -3 dBFS peak, and a sub -60 dBFS noise floor reliably comes from a repeatable chain applied to your edited 24-bit WAV. Step 1, clean up: remove clicks, breaths that distract, and long pauses. Step 2, high-pass filter around 80 Hz to remove rumble and plosive energy that eats headroom. Step 3, noise reduction: capture a noise profile from your room-tone silence and reduce gently (aim to lower the floor below -60 dBFS without introducing watery artifacts — over-reduction sounds worse than mild hiss). Step 4, light compression (roughly 2:1-3:1) to even out delivery so RMS sits in the ACX window without peaks spiking. Step 5, normalize/limit: set a limiter ceiling at -3 dBFS (or lower to -3.5 for safety) and normalize RMS into the -23 to -18 range. Step 6, verify, then export to MP3 192 kbps CBR mono. Build this as a saved chain or macro so every chapter is processed identically.

Do not eyeball ACX compliance — measure it, because automated rejection wastes days. The free ACX Check plugin for Audacity is the standard tool: run it on each finished file and it reports peak level, RMS loudness, and noise floor against ACX limits, flagging pass or fail on each. Adobe Audition users can read the same values from the Amplitude Statistics panel (the Total RMS and peak amplitude fields) and a spectral or noise-floor readout. Check every file, not just one, since a single noisy chapter recorded on a different day can fail. Confirm structural items too: room tone present at head and tail, credits in the right files, mono channel, 44.1 kHz, and 192 kbps CBR in the exported MP3's properties. A quick pre-flight checklist — specs measured, structure confirmed, format verified — turns audiobook submission from a guessing game into a reliable pass.