An audiobook chapter is delivered as one clean file at a consistent level. The chain is trim the mistakes, join the takes, normalise loudness, compress to the delivery limit, then convert to the format the platform wants. Every step runs locally, so an unpublished manuscript is never uploaded.
Order matters because each later step is hard to undo. Trim before joining, since re-cutting inside a joined file means finding the seam again. Join before normalising, so one gain applies to the whole chapter rather than to each take. Normalise before compressing, because compression changes the peaks and the perceived level. Keep the WAV master at every stage and treat the MP3 as the copy you upload. None of this removes noise, so a chapter recorded next to a fridge stays noisy whatever order you work in.
1. Trim the mistakes
Audio Trimmer draws the waveform and lets you drag the region edges or type exact times. It re-encodes the selection rather than copying compressed frames, so the cut lands where you asked, and a fade of up to 5 seconds softens both ends. It opens MP3, WAV, M4A, AAC, OGG, FLAC and Opus, and exports MP3 or WAV. Keep the WAV when the chapter still has to be joined or normalised.
One region per file is the limit, so five fluffs mean five passes. That is annoying but predictable, and it is safer than a multi-cut editor that can leave a glitch at a seam. Do not try to fix room tone here: trimming removes time, not noise. If a take has a hum, a hiss or a chair creak across the whole sentence, the fix belongs in a real audio editor or in a re-record.
Audio Trimmer
Cut a fluff or a pause on the waveform, with a fade of up to 5 seconds.
2. Join the takes
Audio Merger takes the files in the order you set and writes one output, normalising each clip to a shared sample rate and stereo layout first. That is the right tool for three takes that belong back to back, or for a chapter recorded across two sessions. There is no fixed file count, but the job is limited by browser memory.
It is a concatenation, not a mix. There is no crossfade, no overlap and no second track, so it cannot put a music bed under your voice or blend two takes into one. The joins are hard cuts, which is usually fine between paragraphs and noticeable inside a word. Keep the separate takes until the joined file has been listened to end to end.
Merge Audio
Join the takes in order, with one shared sample rate and stereo layout.
3. Normalise the chapter
Volume Booster offers normalisation to -14, -16 or -23 LUFS, a manual gain from -20 to +20 dB in half-step increments, and a limiter that holds peaks at 0.97 so a boost does not clip. Normalisation measures the whole file and applies one gain, which is exactly what makes chapters sit at a similar level after a long recording day.
What it is not is a compressor. It will not pull a loud laugh down against a quiet sentence, and it will not remove a breath. If the performance itself is uneven, that has to be fixed before this step, because one gain cannot reshape a dynamic. Run it per chapter, then check the result with a meter. A platform's loudness target is the platform's rule; this tool can hit a number, but it cannot certify compliance. Output is MP3, WAV or M4A.
Volume Booster
Normalise to -14, -16 or -23 LUFS with a peak limiter at 0.97.
4. Compress to the delivery limit
Compress Audio has two modes. In target size mode you type a size in megabytes, minimum 0.1, and the bitrate is calculated from the length of the file and then clamped between 32 and 320 kbps; the panel shows the rate it actually used. In bitrate mode you choose the rate directly. For spoken word, 128 kbps is transparent enough for most listeners, and a lower rate is fine for a sample or a private review copy.
The output formats are MP3, M4A and OGG, with no WAV, so the result is always lossy. There is no two-pass encode and no way to guarantee an exact byte count, so compress once, download, and check the size the platform reports. Keep the normalised WAV as the master; if a retailer asks for a different specification later, you do not want to start from a lossy file.
Compress Audio
Target a size or a bitrate, with the used rate shown before export.
5. Convert to the delivery format
Audio Converter handles a batch of up to 30 tracks and writes MP3, WAV, M4A, OGG or FLAC. The sample rate can stay as it is, or be set to 44100 or 48000 Hz, and 22050 is available for a small voice file. Converting to a lossless format does not improve a lossy source, so if the master is already a 128 kbps MP3, converting it to WAV only makes it bigger.
Use this step last, once the chapter is final. A delivery folder usually needs one format for the platform and one for your own archive, and running the same finished file through the converter twice is safer than converting, editing and converting again. Keep the original filename and add the format to it, so a chapter, a sample and a master cannot be confused.
Audio Converter
Convert up to 30 tracks to MP3, WAV, M4A, OGG or FLAC.
What this chain cannot do
- No noise removal. Hiss, hum, traffic and mouth clicks stay in the file; there is no denoiser or de-clicker here.
- No mixing. Merge Audio concatenates clips in order; it cannot put music under a voice or blend two tracks.
- No dynamics processing. Normalisation applies one gain and a limiter; there is no compressor, EQ or de-esser.
- No chapter automation. Nothing detects silence, splits a book into chapters or generates chapter markers.
- No platform certification. A loudness number can be reached, but no retailer specification is verified here.
- Compression is one-way and lossy. The compressed output has no WAV option, so keep the master.
- No batch over a whole book. Trimming, merging and normalising work one file at a time; only the converter takes up to 30 tracks.
- Conversion does not add quality. A lossy source stays lossy, and a bigger container does not restore detail.