The 4 steps

  1. Capture — voice notes, typed notes, screenshots of paper notes. Whatever happens when an idea hits you.
  2. Promote — pick one or more related notes, AI drafts a chapter from them into your project.
  3. Package — generate a cover, export the ebook in EPUB / PDF / DOCX.
  4. Narrate — generate the audiobook from the same chapters with the voice of your choice.

Every step lives in one MimicReader account. Notes have a `project_id`, chapters reference their source notes, the EPUB reflects the live chapter list, the audiobook reflects the live EPUB. Edit anything at any step and downstream regenerates from the current state.

Why this matters: the friction between "I have an idea" and "I have a published audiobook" used to be months of context-switching across 5+ tools. With this pipeline it's hours to days, and you never leave the dashboard.

Step 1 · Capture — voice notes, typed notes, anything

Voice notes

Open /notes in the MimicReader app. Hit record. Talk for as long as you need — a 90-second walking thought, a 20-minute scene description, an interview transcript. Whisper Large V3 Turbo transcribes locally to markdown.

The transcript becomes a markdown note in your Inbox. You can edit it, tag it to a project, link it to a chapter, or just leave it as a loose thought to come back to later.

Typed notes

Same Notes tab, click New note, type. Plain markdown editor — `**bold**`, `_italic_`, `## Section`, bullet lists with `-`. Notes saved continuously.

AI reformat

Got a wall-of-text dictation that's hard to read? Click Reformat with AI. The AI cleans up filler words ("um", "y'know"), splits run-on sentences, adds light structure (paragraphs, bullets, headings where appropriate). Your original is preserved — reformat is non-destructive.

Project binding

Each note can be tagged with a Project. Tagged notes show up in the project sidebar. Untagged notes live in Inbox. You can mass-move from Inbox to a Project later, so don't worry about tagging during the capture step — capture first, organize later.

Start capturing free

Voice notes, typed notes, AI reformat — included in every free account. 1 hour of audiobook generation per month, no credit card needed.

Open MimicReader

Step 2 · Promote — turn notes into chapter drafts

This is the step that makes the rest of the pipeline cheap. Without it, you'd have to manually compose chapters from your notes. With it, the AI does the first draft, you edit.

How it works:

  1. Open a note in your project
  2. Click Promote to chapter
  3. The AI Workshop reads the note(s), considers project context (genre, target audience, language), drafts a chapter in your project's tone
  4. The new chapter appears in the project sidebar; the source note links to the chapter so you can trace where ideas came from

You can promote multiple related notes into a single chapter. The AI synthesizes — it doesn't just concatenate. If your three notes are about the same scene from three angles, you get one coherent scene out, not three pasted blocks.

What the AI Workshop does and doesn't do

Workshop templates handle structural AI tasks: chapter drafting, synopsis generation, character name suggestion, blurb writing. It does not:

Use it for momentum. Edit the output. The first-draft-from-notes flow saves hours per chapter, even if you rewrite half the result.

Skip Promote if you want

The pipeline works without it. Just write chapters directly in the Project editor and skip step 2. Promote-to-chapter is for the notes-first workflow specifically.

Step 3 · Package — ebook, cover, exports

Once your project has chapters (whether AI-drafted or hand-written), click Generate cover. Pick a style:

StyleBest for
Oil paintingLiterary fiction, historical, classic
Fantasy illustrationFantasy, sci-fi, YA adventure
Minimal modernNon-fiction, business, design
WatercolorMemoir, poetry, children's
PhotorealisticThriller, contemporary, crime

Flux generates the cover from your title and blurb (~30 seconds). If you don't like it, regenerate. You can also upload a custom cover if you already have one.

Then click Export:

The EPUB includes your cover, title page, table of contents, properly chunked chapters with semantic HTML, and NCX/NAV navigation that works on every major e-reader.

Skip if you only want audio: you don't have to export the EPUB before generating the audiobook. The audiobook generator reads chapters directly from your project. Export is for distribution, not a pipeline dependency.

Step 4 · Narrate — generate the audiobook

Same project, click Generate audiobook. Pick:

The TTS pipeline:

  1. Concatenates chapters into one long text with proper pause markers at chapter and paragraph boundaries
  2. Detects dialogues and applies dialog-specific narration parameters (higher exaggeration, dialogue-style cadence)
  3. Detects emotional words and adjusts intonation per chunk (anger, joy, sadness, fear, etc.)
  4. Generates in parallel across 3 GPU workers (real-time factor ~0.3 — 1 hour of audio in ~18 minutes wall time)
  5. Cross-fades, normalizes to EBU R128 broadcast standard, encodes M4A with chapter markers + sync.vtt for read-along

For a 60,000-word novel: ~6 hours of audio, ~110 minutes of generation. Free 1-hour monthly credit covers a novella; the £2.50/month subscription gives 2.5 hours/month; PAYG packs (£5, £20, £100) for one-off bigger projects.

Real workflows in the wild

The casual writer (a friend wrote a memoir this way)

Records 15-20 minute voice memos during morning walks. Each memo becomes a note. At the end of the week, picks the 4-5 memos that belong to "Childhood — moving to Glasgow" and clicks Promote — gets a chapter draft. Edits in the manuscript editor. After 3 months: 18 chapters, exports EPUB for family, generates audiobook for older relatives who prefer listening. Total cost: £0 (everything inside free tier + one £5 pack for the full audiobook).

The researcher (humanities PhD)

Interviews 30 subjects, uploads each interview as a voice note (Whisper transcribes). Tags each by theme. Uses Promote to convert clusters of interviews into thematic chapters. Exports DOCX to share with their supervisor. Final publication: EPUB + audio version (academic library prefers audio for accessibility compliance).

The indie author (NaNoWriMo type)

Writes 50,000 words in November as plain Project chapters. Skips notes step entirely. Generates cover (fantasy style) at the end of November. Exports EPUB and uploads to KDP. Generates audiobook in December. Lists on ACX as royalty-share or paid distribution.

The educator (corporate training)

Captures lectures as voice notes (Whisper transcribes ~95% accurate). Promotes to chapters with the Workshop "summarize and structure" template (still in development; today: manual organization). Exports as PDF for a study guide and as audiobook for the commute. Distributes via internal share links.

What's intentionally not part of the pipeline

Honest disclosure of friction points:

Try the pipeline now

Every account gets the full 4-step pipeline:

From idea to audiobook — same dashboard

Capture, draft, package, narrate. The pipeline that doesn't exist anywhere else.

Start Free

Where to go next