Article to Audio: Turn Long Reads Into MP3s to Play

Get clean text before you convert an article to audio

The voice reads everything you paste, so start with only the story itself.

• Open the page in reader mode: Safari Reader, Firefox Reader View or Edge Immersive Reader strip menus, ads and related links. Then select all and copy.
• Delete what is meant for eyes: image captions, pull quotes that repeat the body, newsletter sign-up boxes, share prompts and footnote markers.
• Replace URLs with a spoken cue such as link in the original, or remove them; a voice reading a long web address is painful.
• Add the headline and author as a first line, so the file announces itself when you open it weeks later.
• For PDFs such as reports or papers, pull the text out first with the PDF to Text tool, then fix hyphenated line breaks.

Two minutes of cleanup is worth it: a stray caption in the middle of a paragraph breaks the flow more than any voice quality issue, and a clean copy also gives you a tidy text file to search later when you want to quote a passage.

Splitting long reads into listenable parts

The text box holds 20,000 characters, roughly 3,000 English words or about 20 minutes of speech. Longer features need to be split, and good split points make the result nicer to listen to.

• Cut at section headings or scene breaks, never mid-paragraph.
• Name files in order, such as atlantic-heat-01.mp3, atlantic-heat-02.mp3, so players sort them correctly.
• Start each part with a short cue, such as Part two, so you know where you are when you resume.
• Inside each run, the tool itself breaks text into sentence-sized pieces and shows progress as part 3 of 40, then joins them with short natural pauses.
• If you only need the gist, summarize first with an AI summarizer and convert the summary; note that AI writing tools run in the cloud, unlike the voice model.

Generation happens on your device, so speed depends on its processor. A laptop handles a 20-minute part comfortably; an older phone can take a while, so start the long ones plugged in.

Choosing a voice and speed for long listening

A voice that sounds fine for 30 seconds can grate after 30 minutes. Pick for comfort rather than drama.

• Warm, mid-pitched voices wear best over long sessions: try Heart or Sarah (US female), Michael or Eric (US male), Emma or George (UK).
• Avoid the deepest voice for news and analysis; Onyx suits trailers better than a 40-minute read.
• Generate at 1.0x and speed up in your player later if you like; most phone players offer 1.25x to 2x without changing pitch.
• Use 0.8x to 0.9x for dense material or articles in a language you are still learning.
• Keep one favorite voice for your whole reading queue; consistency makes switching between articles effortless.

The downloadable voices are English only. To hear an article in another language, switch to Device voices, which play instantly in dozens of languages but cannot be saved as files.

Getting the audio onto your phone

Each part downloads as a 128 kbps mono MP3, about 1 MB per minute, or as a lossless WAV about three times larger. MP3 is the right choice for listening.

• iPhone: AirDrop the files or save them to iCloud Drive, then play from the Files app or a player such as VLC, which remembers your position.
• Android: copy to the Music or Download folder over USB, or save from Google Drive for offline use, and play in any audio app.
• Make a playlist in your player named Reading queue and add new parts as you make them.
• Delete parts you have finished; a month of daily long reads at 30 minutes each adds up to roughly 900 MB, which matters on a 64 GB phone.

Keep it personal. Converting an article for your own listening is a common fair use, but publishing or sharing narrated copies of paywalled or copyrighted pieces is not, so do not post them as podcasts without permission from the publisher.

Step-by-step

Common mistakes to avoid

Pro tips

Frequently asked questions