← Back to the app
Book reading session · about one hour
This recording does two jobs: it trains your own voice model, and it is the pilot for the audiobook. Read it the way you would want to hear it: unhurried, conversational, talking to one reader.
The plan · read in this order
| File | Section | Time |
|---|
| Book 00 | Introduction | ≈ 10 min |
| Book 01 | Chapter 1 · The Night Money Changed | ≈ 12 min |
| Book 02 | Chapter 2 · The Cure That Can Never Be Given Again | ≈ 14 min |
| Book 03 | Chapter 3 · The Put | ≈ 14 min |
| Book 14 | Chapter 14 · The Choice | ≈ 9 min |
Skip the draft notes in square brackets at the top of each chapter, headings are optional. If you run out of steam, stop after any chapter: three chapters already make a good model.
Three rules that matter
- One file per chapter. Record each chapter as its own file, name it
Book 00, Book 01, … Pause the recorder between chapters, not inside them.
- Same setup for the whole hour. Same room, same distance (a hand's width from the phone), same volume. Do not move to another room half way. If something interrupts, just carry on, a flubbed line is fine: re-read the sentence and keep going. Do not stop and restart the file.
- Your natural low register. Your real voice sits low; every cloud clone so far pitched you up. Read as you speak, not as a presenter. Drink water before, not during.
Phone settings
- Voice Memos (iPhone) or Recorder (Android), compressed / AAC is fine. Lossless makes files too big for upload; the model trains on 24 kHz anyway.
- Airplane mode on, notifications off, no fan or fridge in the room, soft furnishings around you (a closet full of clothes is ideal).
- Each chapter file should be under 25 MB: AAC at 15 minutes is about 14 MB.
After recording
Open the app → Voice → Upload a file, one chapter at a time, label exactly Book 00 … Book 14. The studio machine measures each file (noise, clipping, levels) within a few minutes and shows a verdict next to it. Then say the word and the training run starts.