Just say it.

Hold a key. Say who and what. It's there.

P

To Pulkit

Hold Space and speak

|

↵ send · esc cancel

Release. Glance. Press Enter. You never leave your app.

Build from source See how it works

Your voice stays on this device. Zero network calls. No log.

Writing Hold Speak Words appear ✓ Sent

The cost of switching

You leave your work to send one line.

You switch apps, find the chat, type, get pulled into other chats. You lose where you were.

Before

0s

Switch app, find chat, type, get distracted, find your place again.

With Voicy

0s

Hold. Say who and what. Release.

How it hears

Your sentence splits in one place.

Before the split is who. After is what you said, word for word.

message Pulkit that I'm on my way
dropped who dropped kept exactly

The last part is copied by position. No rewrite.

Four things we will not do

Promises

Keep your words

Your words stay exactly as you said. No rewriting. Your voice stays yours.

Ask, not guess

Two people called Pulkit? We ask which one. Wrong send breaks trust.

Stay on this device

Turns voice to text here. Contacts stay here. Zero uploads.

Works where you type

Same hold and say in any app. No switching, no paste dance.

Heard as “Paul Kit”. Still finds Pulkit.

We tried nudging the listener. Same words back. So we fix it after, by sound and spelling. Correct once, we remember.

“Paul Kit”

Pulkit

One correction lasts forever.

Try it

Two minutes.

  1. 1 Clone and build
  2. 2 Hold Ctrl Space, say a name and a message
git clone https://github.com/Rishet11/voicy.git
cd voicy
./build.sh
open dist/Voicy.app

Needs desktop app installed where you send messages. Check: ./dist/Voicy.app/Contents/MacOS/Voicy --selftest

After you open it

  1. Find voicy in menu bar
  2. Allow Microphone and Contacts
  3. Hold, speak, release, press Enter

Works where you type. You stay in your app.

Questions.

Will it get me banned?

No. We open the app with the link it publishes for the web. No private API.

Does audio leave this device?

Never. It runs here. We make zero network calls and store no messages.

What if it hears the name wrong?

It will. We match by sound and spelling. Teach it once and it sticks.

Why not fix my words?

Because your voice is yours. We copy. We do not rewrite.

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

Read the source Star on GitHub