Dictation for Mac, Windows and Android, with your own AI

Hold a key.Say it.It's written.

SayFeather turns what you say into clean running text, right where your cursor is. In any app. Speech recognition runs on your own device; the clean-up is done by the AI provider you connect yourself.

Free to use. The apps update themselves.

Three steps, a second or two.

1

Hold your key

Pick any hotkey, or the fn key on a Mac. Hold it while you talk, or tap once to start and once to stop.

2

Say what you mean

Talk the way you talk. Whisper turns your voice into text on your own device; your audio never leaves it.

3

Let go

The text lands at your cursor: mail, chat, documents, a terminal. No window to switch to, nothing to copy.

You talk like a person. It writes like you meant it.

Connect an AI provider and SayFeather tidies each dictation: fillers and repeats go, sentences run on properly. What you said stays what you said.

As heard

uhm so hi Sanne can we can we move the meeting on Thursday to uh to Friday morning I have an appointment that I can't I can't change

Cleaned up

Hi Sanne, can we move Thursday's meeting to Friday morning? I have an appointment I can't change.

The clean-up only corrects. It never rewrites your message and never answers it.

Built for people who type all day.

Works in every app

Wherever a cursor blinks, SayFeather can type. No plug-ins, no separate editor.

Speech stays on your device

The Whisper speech model runs locally. It downloads once, then works without an internet connection.

Your own AI provider

Use the provider and key you already have. The app calls it directly, so SayFeather never sees your text.

Enter, when you want it

Send straight away in terminals and AI chats, not in mail or documents. Set it per app, and flip it once with a key.

Fully offline, if you like

Pair it with a local model through Ollama or LM Studio and nothing leaves your computer at all.

On your phone too

On Android, SayFeather is a keyboard, with a floating button that works above any other keyboard.

Connect your own AI

SayFeather has no AI subscription of its own. You pick a provider and model, paste your API key into the app and pay that provider directly. Or run a local model and pay nothing.

Have Claude Max or Ultra?Your plan includes $100 (Max) or $200 (Ultra) of Claude API credit every month. Use it for SayFeather at no extra cost.How to use your Claude creditConnects to
  • Anthropic
  • OpenAI
  • Gemini
  • Mistral
  • Groq
  • OpenRouter
  • Ollama
  • LM Studio
  • Any OpenAI-compatible address

Your words are yours.

SayFeather has no server that hears you or reads along. That is how it is built, not a setting you have to find.

  • Audio never leaves your device. Speech recognition runs locally.
  • Text goes only where you send it. The clean-up goes straight from your device to the provider you chose, with your key.
  • Our service only checks your licence. It never receives what you dictate.

For businesses

Let your users talk to your own app.

The engine behind SayFeather is a separate service: audio in, clean text out. Build it into your product, so your customers speak instead of type.

A microphone button

A small component you drop into your web app or dashboard. Your users hold it, talk, and the text lands in the field.

Runs where you want

On your own servers or close to your users, with the speech model and language model you choose.

Your data stays yours

No audio or text passes through SayFeather. You decide where it goes.

In development

We are building this now. Tell us what you want to use it for, and help shape it as an early partner.

Get in touch [CONTACT]

Free to dictate. Pay once for the clean-up.

No subscription. You pay your AI provider for what you use, and nothing to us per month.

Coming from Wispr Flow or another app? See what you save.

Free

€0

The text as you said it.

  • Dictate in every app
  • Local speech recognition
  • Enter after dictating, per app
Download

Questions

Does my voice go to a server?

No. Your voice becomes text on your own device. Only if you connect an AI provider does the text, never the audio, go to that provider for the clean-up.

What does the clean-up change?

It removes fillers and repeats and makes running sentences. It does not change what you said and never answers a question you dictate.

What can I do without a licence?

Everything except the clean-up. SayFeather types the text exactly as it heard it.

Is it a subscription?

No. The licence is a single payment. Your AI provider bills you for your own use, if it charges at all.

Is there an iPhone app?

Not yet. An iPhone keyboard is not allowed to use the microphone, so that needs a different approach.

Stop typing what you could just say.

Free to download. The apps update themselves.