Why offer audio
- Readers listen on the move: commuting, cooking, walking the dog.
- Accessibility for readers with visual impairments, dyslexia or tired eyes.
- The writer's own voice feels personal in a way a generic reading voice doesn't.
- A podcast for free: an audio edition of a newsletter is a ready-made show.
A weekly routine
- Once: clone your voice from a 10-second sample in your natural reading voice. See cloning your own voice.
- Once: make a project for the publication.
- Each issue: add a script, paste the piece with a blank line between paragraphs, and each paragraph becomes a line.
- Tidy it for the ear: replace "click here" and link text, write out numbers and abbreviations, describe charts in a sentence.
- Generate all new lines, listen through, re-take anything misread, save.
- Download, join the paragraphs into one file with short pauses, and publish.
Adapting written pieces for listening
- Headings become short lines on their own: "Part two. What the data says."
- Links: drop "here" and "this link"; say the source instead: "according to a recent report from the city council".
- Lists become sentences or short lines.
- Images and charts: one sentence on what they show, or leave them out.
- Quotations: read by the narrator ("She told me, and I quote...") or given a second voice, from someone who agreed.
Long pieces
Each line can hold up to 2,000 characters, and long lines are made a few sentences at a time and joined seamlessly, but a paragraph per line makes corrections quick. See narrating long text.
Publishing
- Master to about -16 LUFS for podcast feeds. See loudness standards.
- Export MP3 for the web; keep the WAV as your master.
- Tell listeners it's an AI version of your voice, for example at the top of the episode notes.
Guest writers
If you publish guest pieces, don't read them in a clone of the guest's voice without their written permission. Your own voice, or a house narrator, is the safe default. See the consent template.