ACX Audiobook

Each file must sit inside a loudness range, stay under a noise floor, keep peaks below a ceiling, and carry a specific amount of room tone at the head and tail. One file per chapter, consistent across the whole book.

Three of those four are fixable after recording with normal editing. The noise floor is the one that is not, because you cannot remove a room you have already recorded without the voice going with it.

A cheap microphone in a quiet soft room beats an expensive one in a hard empty one, every time. Carpet, curtains, a wardrobe of clothes, a duvet behind you. Turn off the fridge, the boiler and anything with a fan.

Record thirty seconds of silence in your space and listen to it at high volume. Whatever you hear is on every page of your book.

Finished audio runs at roughly a third of your recording time, so an eight-hour book is about twenty-four hours in the chair. Two-hour sessions, water at room temperature, and stop when your voice tires rather than pushing through, because tired voice is audible and unfixable.

When you fluff a line, pause, clap once, and read the sentence again from the start. The clap shows as a spike in the waveform and makes the edit findable without listening back.

Measure loudness across the whole file, not a section. Check the peak. Measure the noise floor in the room tone at the head. Confirm the head and tail room tone lengths. Then listen to the first and last thirty seconds of every chapter, which catches most human errors.

Noise floor too high, which means the room. Loudness inconsistent between chapters, which means the setup moved. Missing or wrong room tone at head and tail. And an opening or closing credit that does not match what the platform expects.

Not one of them is about the quality of the narration. They are all production, which is good news, because production is learnable in an afternoon.