← Back to blog Workflow · 5 min read

Turn a Substack archive into an audiobook in one afternoon.

May 2, 2026

You've got 50 essays, the kind that take 4-6 minutes each to read, and people keep asking for an audio version. Until recently the answer was: book a studio, hire a narrator, mix in Pro Tools, master to ACX. Or just don't.

Here's how to do it on Artistik in the span of an espresso and a walk around the block.

The five steps

1. Export your Substack archive. Settings → Exports → Download. You'll get a ZIP with one HTML file per post. Pick the 20–30 essays you want in the audiobook. Concatenate them into a single document — Pages, Word, anything. Add a quick foreword and a one-paragraph intro per essay. Rough word count after pruning: 30k–60k.
2. Open the studio and create an Audiobook job. Drop the document into the upload zone. Artistik extracts the text — no copy-paste, no formatting drift.
3. Pick a voice. The agent presents 4–6 ElevenLabs voices that match your tone brief (warm and conversational? newsroom-precise? story-time intimate?). Each plays a 5-second sample. Confirm the one you want.
4. Walk away. The agent renders chapter by chapter, watches loudness, retries any clip that comes out muddy. You get an email when the full-book MP3 is ready.
5. Ship it. Upload the MP3 to ACX, Findaway Voices, Apple Books, or just drop it in your Substack as a paid-tier perk.

What it costs

ElevenLabs is the only meaningful cost — about $0.00012 per character. A 50k-word essay collection is roughly 280k characters, so the raw API cost lands around $33. With Artistik's 1.4× Pro markup, you're paying about $46. Plus an EPUB pass if you want the text edition too.

That's a tenth of what a human narration session costs, and you can iterate — try a different voice, try a different pacing — without booking another studio day.

What it doesn't do

Some things still need human judgment:

• Pronunciation of names and proper nouns. ElevenLabs gets it right ~95% of the time. The other 5% you'll catch on a listen-through and re-render the affected line.
• Emotional inflection on a punchline. The narration is good, but it's not Audible-narrator-of-the-year good.
• Music + sound design. Out of scope for the audiobook agent — drop the MP3 into a DAW if you want bumpers.