On-device neural voices
Choose from five model families — Supertonic, Kitten, Kokoro, Piper, and Tiny TTS. Supertonic ships bundled; other models install from Hugging Face, cache locally, and synthesize in-browser.
Flow TTS
Text-to-Speech Chrome extension for listening to web pages
Flow TTS is a text-to-speech Chrome extension that reads web pages aloud — privately, on your computer. Natural voices run on-device for supported local models: WebGPU when it's there, WebAssembly when it isn't. Use Auto mode to read the main text on a page, or Flow mode with Flow Templates that remember the readable sections you choose on each site.
- Auto mode for the main text, or Flow Templates you save
- On-device speech generation for supported local models
Everything you need, nothing you don't
A focused set of tools designed to make reading with your ears effortless.
Two reading modes
Auto mode reads the main text on the page. Flow mode plays the Flow Template you saved for that site. Switch between them in one click.
Flow Templates
Pin elements on any page into a reusable listen plan. Flow TTS remembers the structure and uses it again on later visits.
Streaming playback
Audio begins playing the moment the first chunk is ready — no waiting for the entire sentence to finish generating.
Fully configurable
Speed, volume, voice, sample rate, inference quality, player position, pronunciation rules — everything is adjustable.
Private by design
Speech generation runs on-device for supported local models. Account setup, usage checks, license status, and model downloads may use network services. Optional diagnostics are opt-in only.
See it on the page
The floating player, Flow Templates, and voice picker — the same UI you get after install.
Up and running in seconds
Install, create your free account, choose a voice, and press play.
1
Install
Add Flow TTS from the Chrome Web Store. Pin it so the player is one click away.
2
Sign in
Enter your email and the one-time code. You can tour first; playback needs a signed-in account.
3
Open on any page
Click the toolbar icon to summon the floating read controller directly on the page.
4
Press play
Auto mode reads the main text. Switch to Flow mode when you have a Flow Template for that site.
The engine behind the voice
A library of neural models runs on-device for supported local models — download from Hugging Face, swap at will, and playback begins while the next sentence is still generating.
This page runs a live check in your browser using navigator.gpu.requestAdapter(). That is the same probe Flow TTS uses before enabling the WebGPU inference path (otherwise it falls back to WASM).
On-device synthesis
ONNX neural models run locally via WebGPU or WASM for supported installed models. Speech generation stays on-device; model downloads and account checks may use network services.
Streaming playback
Audio starts playing as soon as the first chunk is synthesized — no waiting for the full sentence to render.
Fine-tune everything
Speed, volume, voice, sample rate, and inference quality steps — all adjustable per your preference.
Smart text pipeline
Strips emojis, normalizes punctuation, handles abbreviations, and chunks text at natural sentence boundaries.
Flow Templates
Think of a Flow Template as a listen plan for your favorite sites. Tell it what to read and what to skip — it remembers for next time.
Visual element picker
Point-and-click any element on the live page. Mark it Read, Skip, or Repeat similar matches — no code required.
Live-root reading
Pick a region, then choose Normal, Dynamic extract, or Read everything inside so changing pages stay in the plan.
Reusable Flow Templates
Save a listen plan per site. Come back later and Flow mode finds those same structures and plays them again.
Power user controls
URL scoping, read-in-page-order, repeatable sections, up/down reorder, and manual selector overrides for precise control.
A voice library, on demand
Supertonic ships bundled so you can start listening immediately. Install Kitten, Kokoro, Piper, Tiny TTS, or Supertonic v2 from Hugging Face in Settings → Models. Speech generation runs on-device for supported local models.
Built-in TTS model manager
NewBrowse installable models, fetch the weights from Hugging Face, and Flow TTS handles caching, swapping, and WebGPU/WASM routing. Finish an install before you play that model.
Kitten
FastestInstall-only Kitten v0.8 variants. CPU-only via WASM — skips GPU initialization. Fast starts on lighter machines.
- Pure WASM, zero GPU overhead
- Install from Settings → Models
- Tiny footprint (~25 MB)
Piper
LightweightCPU-optimized ONNX runtime with a broad library of community-trained voices and languages. Graceful on modest hardware.
- Wide language coverage
- Large community voice catalog
- Low memory footprint
Supertonic
High fidelityWebGPU-accelerated when available, with a clean WASM fallback. Best when voice quality matters most.
- Bundled and ready — no download
- WebGPU + WASM fallback
- Configurable inference steps
Kokoro
ExpressiveAn 82M-parameter StyleTTS2-based model with a rich library of expressive voices. Runs on CPU or WebGPU.
- 82M parameters
- Expressive multi-voice library
- WebGPU or WASM inference
Private by design
Speech generation runs on-device for supported local models. Account setup, usage checks, license status, and model downloads may use network services.
On-device synthesisLocal model supportOptional diagnostics onlyChrome MV3
Plans
Start free after account setup. Upgrade on Lemon Squeezy — Monthly, Yearly, or Lifetime.
Monthly
$6.99
$6.99 billed every month
- Unlimited characters — no daily cap
- Start this month without a year-long lock-in
- Cancel whenever you're done
Yearly
2 months free$69.00
$69.00 billed every year
- Unlimited characters for the whole year
- Ten months' price for twelve
- One checkout, then get back to reading
Lifetime
Pay once$149.00
Pay once, no renewals
- Unlimited characters with no renewal dates
- One payment and billing is done
- Keep Premium without a subscription calendar
Every paid plan includes everything on the Free plan.
Free
$0after sign-in
Core reading after a free account. Enough to try Auto mode, Flow Templates, and on-device voices before you pick a paid plan.
- Auto mode — reads the main text on the page
- Flow mode with reusable Flow Templates
- Five TTS model families and the model manager
- Custom pronunciation, player docking, and hover-to-play
- On-device speech for supported local models
- 10,000 characters per day
Common questions
How Flow TTS is installed, priced, and different from online generators or dictation apps.
Is Flow TTS free?
Yes. After you create a free account with an emailed one-time code, you can use core reading features with a 10,000-character daily cap. Premium — Monthly, Yearly, or Lifetime — removes that cap and unlocks Glass Mode.
How do I install Flow TTS?
Install it from the Chrome Web Store. There is no desktop app or standalone installer. After you add the extension, pin the toolbar icon, then sign in with your email and one-time code to start playback.
Does Flow TTS work on Android?
No. Flow TTS is a Chrome extension for desktop Chrome. Chrome Web Store extensions do not install on Android Chrome. Use it on a computer, or pick a mobile reader if you need listening on a phone.
Is Flow TTS an online text-to-speech website?
No. Flow TTS is a Chrome extension that reads the page you are already on. You do not paste text into this website to generate audio. Open an article, click the toolbar icon, and press play.
Is Flow TTS an AI voice generator?
It uses neural voices to read web pages aloud, on-device for supported models. It is not a paste-text AI voice generator. You listen to the current page rather than typing a script to produce a clip.
Is Flow TTS the same as Flow AI or a dictation app?
No. Flow TTS is text-to-speech: it reads page text aloud. Dictation apps and other products named Flow turn your speech into text. Flow TTS does not transcribe your voice.
Start listening.
Add Flow TTS from the Chrome Web Store — free account, on-device speech for supported models, and built for how you actually browse.