Reading · checked 2026-10-13

Can I vibecode Speechify?

YES· one-shottable

What people do in Speechify every day is simple: open a PDF, a web page or a book, press play, and follow the highlighted text at a speed they choose. Kokoro, a free open voice model, sounds close to the paid voices and runs on an ordinary laptop, so an agent can build that reader in one sitting. What you give up is the phone apps, the camera scanner for paper pages and the licensed celebrity voices. Exporting chapters as MP3 files covers most listening away from the desk.

$29/moprice

$348you'd save a year

one sittingbuild time

—people replaced it

Before you start

What you need on hand.

DifficultyIntermediate · Some system setup or keys.

TimeAbout an hour

Claude Code, Codex or CursorAny of them can run the prompt.
Python 3.11 and uvThe Kokoro model runs on CPU; Apple Silicon or an NVIDIA card makes it faster.
espeak-ng, tesseract and ffmpegOne Homebrew or apt command; espeak-ng helps Kokoro pronounce unusual words, tesseract reads scans, ffmpeg makes the MP3s.
About 1 GB of free diskFor the voice model and cached audio.

Before you buildSpeechify's free plan only has robotic voices, but the Read Aloud built into Microsoft Edge uses natural voices at no cost. Try it before you build.

1. Install one agent: Claude Code, Codex, Cursor.
2. Make an empty folder and open it in the agent.
3. Paste the prompt below, press enter, and answer its questions as it builds.

The prompt

Build me a local read-aloud app to replace Speechify. Requirements:

- Python 3.11 + FastAPI with a plain HTML/JS front end on
  http://localhost:5050; `uv run app.py` starts everything.
- Voices: Kokoro-82M through the `kokoro` package, running locally on CPU or
  Apple Silicon. List every voice it ships; if OPENAI_API_KEY is set in .env,
  offer OpenAI's TTS voices as well.
- Input: pasted text, a URL (main text extracted with trafilatura), PDF
  (PyMuPDF, dropping repeated headers, footers and page numbers), EPUB
  (ebooklib), DOCX (python-docx), and photos or scanned PDFs via tesseract OCR.
- Reader: split into sentences, synthesise 3 sentences ahead, highlight the one
  being read, click any sentence to jump there. Speed 0.75x to 4x with pitch
  kept natural (preservesPitch), a voice picker, space to pause, arrows to skip.
- Library: SQLite (stdlib sqlite3) at ~/ReadAloud/library.db with each document,
  its source and my last position; reopening resumes where I stopped.
- Cache audio per sentence and voice in ~/ReadAloud/cache/ so re-reading is free.
- Export any document to MP3, or to M4B with one chapter per section, using
  ffmpeg into ~/ReadAloud/exports/, so I can listen on my phone.
- No accounts, no telemetry, local-first: nothing leaves the machine unless I
  add an API key.
- Out of scope: mobile apps, a browser extension and AI summaries.
- README: installing espeak-ng, tesseract and ffmpeg, the first-run model
  download (about 350 MB), and how long a chapter takes on a CPU-only laptop.

Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.

Share on X

Max MiskowSponsoredAI agents, automations and MVPs, built for your company.Visit site ›

Then keep going

Next prompts, same folder.

The first prompt builds the core. Run these one at a time in the same folder to grow it into your own Speechify.

Add a Chrome extension (Manifest V3) to the read-aloud app in this folder. Requirements:

- A toolbar button and a right-click menu item: "Read this page" and
  "Read selection".
- Send the page URL or the selected text to the local server on
  http://localhost:5050 with a token generated by the server and kept in .env.
- The server adds it to the library and opens the reader tab at the start.
- If the server is not running, the popup says so.
- Update the README with how to load the unpacked extension.

Turn the exports of the read-aloud app in this folder into a private podcast feed. Requirements:

- Serve an RSS 2.0 feed with iTunes tags at /feed/<token>.xml listing every
  MP3 or M4B in ~/ReadAloud/exports/, newest first, with title and duration.
- The token is random, stored in .env, and the files are only served with it.
- Bind to 0.0.0.0 only when FEED_PUBLIC=true in .env.
- Update the README with how to reach the feed from a phone over Tailscale
  and add it in Overcast, Pocket Casts or Apple Podcasts.

Add a phone scanning page to the read-aloud app in this folder. Requirements:

- A mobile-friendly page at /scan with a camera input
  (<input type="file" accept="image/*" capture="environment">) that takes
  several photos in a row.
- Straighten and sharpen each photo with OpenCV, run tesseract OCR, and join
  the pages into one document in the library.
- Show the recognised text for a quick check before saving.
- Update the README with how to open it from a phone over Tailscale (HTTPS is
  needed for the camera).

Add chapter summaries to the read-aloud app in this folder. Requirements:

- A Summarise button on each document sends its text, one chapter at a time,
  to the Anthropic API (ANTHROPIC_API_KEY in .env).
- Store a 5-bullet summary per chapter in SQLite and show it above the text.
- A "Listen to summary" button reads the bullets with the current voice.
- Never send a document unless I press the button.
- Update the README with what leaves the machine.

What you lose

iPhone, Android and Mac apps with your library and place synced between them
Scanning paper pages with the phone camera
Over a thousand voices, including licensed celebrity ones
The Chrome extension that reads any page or Google Doc
AI summaries, AI podcasts and voice typing

Why people still pay

Many of its users listen because reading is hard work for them, and Speechify is even sold through disability support schemes such as the UK's DSA and Access to Work. For someone who listens every day, a finished app on every device, with the place in each document kept in sync, is worth more than the fee, and the camera scanner and familiar voices are hard to give up.

Moatexecution polish content & rights

Do the math

Speechify or your own copy?

Over

Your usage

Count my time3h ×$/h

Speechify, 1 yr$348

Your own copy, 1 yr$17running + build session

You keep$331pays for itself in the first month

Estimate · Kokoro runs on your machine, so listening costs nothing. An optional cloud voice through an API key costs cents per chapter. The build is token usage on your agent plan.

Leaving Speechify

Take your data, then cancel.

1 · Take your data out of Speechify

1Your library mostly holds copies, so keep the original PDFs, EPUBs and documents, or fetch them again from Google Drive, Dropbox or OneDrive where they live.
2Speechify documents no bulk export of the library: gather the originals of anything you uploaded only to Speechify before you cancel.

2 · Cancel Speechify

1Bought on the web: log in at app.speechify.com, click your photo in the top right, open Settings and choose Cancel Subscription in the left panel.
2Bought on iPhone or iPad: cancel in your Apple Account under subscriptions; Speechify can't do it for you.
3Bought on Android: cancel in Google Play under Subscriptions.

Trials turn into a paid plan on their own and yearly plans renew at the current price, so cancel well before the date; yearly subscribers get a reminder email 15 to 45 days ahead.
Premium stays on until the paid period ends, and unused days are not refunded. App Store and Google Play purchases are final as far as Speechify is concerned.
If you need help, write to Speechify support with your account email and where you paid.

Checked 2026-10-13 against Speechify's docs: 1, 2, 3, 4

The escape hatch

Don't feel like building it? These are already free.

OpenReaderOpen sourceSelf-hosted read-along reader for EPUB, PDF and DOCX with word-by-word highlighting and MP3 or M4B export; plug in Kokoro or a cloud voice.★ 540 · active Oct 2026 · docker to self-host
WebVisit ↗ReadestOpen sourceE-book reader for every platform with built-in read-aloud, sync and notes.★ 25k · active Oct 2026 · one-click install
macOS, Windows, Linux, iOS, Android, WebVisit ↗abogenOpen sourceTurns EPUB, PDF and text files into audiobooks with Kokoro voices and matching subtitles; Windows gets a script installer.★ 6.1k · active Oct 2026 · one-click install
Windows, macOS, LinuxVisit ↗

All 7 free alternatives to Speechify ›

The numbers

Speechify pricing

Plan	Monthly	Annual, per month	What you get
Free	$0/user	$0/user	10 robotic-sounding voices, speeds up to 1.5x, text-to-speech only.
Premium	$29/user	$11.58/user	1,000+ voices in 60+ languages, up to 5x speed, Scan & Listen, AI summaries and chat, Google Drive, Dropbox and OneDrive imports, voice typing, AI podcasts. Premium voices are guaranteed for at least 150,000 words a month.

Free tier: Only robotic voices, no faster than 1.5x, and none of the scanning or AI features.
Billing: Monthly ($29) or yearly ($139); renews automatically at the then-current price, with a reminder email before yearly renewals. Trials (3 days by default) turn into a paid plan unless cancelled. No prorated refunds for the current term.
Hidden costs: Premium voices have a monthly word allowance (150,000 words guaranteed, with a higher promise on the usage-limits page for 2026); past it, Speechify may restrict premium voices or sell extra words. App Store and Google Play purchases are final with Speechify and must be cancelled in those stores. Speechify Studio and the API are priced separately.
Source: speechify.com

Also on the list

ElevenLabs$6/mo · AI audioKINDA

Wispr Flow$15/mo · DictationYES

Airtable$24/mo · DatabasesYES

Questions

Speechify, briefly

Not answered here? Tell us and we'll add it.

Ask a question ›

The free plan reads aloud with 10 robotic voices at up to 1.5x speed.

Yes. An AI coding agent such as Claude Code, Codex or Cursor can build a usable personal version in one session with the prompt on this page, running on your own machine.

About $29 a month on the Premium plan, $348 a year (checked 2026-10-13).

iPhone, Android and Mac apps with your library and place synced between them; Scanning paper pages with the phone camera; Over a thousand voices, including licensed celebrity ones; The Chrome extension that reads any page or Google Doc; AI summaries, AI podcasts and voice typing. If any of those matter to you, keep paying.