Editorial concept art of three people speaking into microphones

A faster way from thought to text

Your voice has momentum.

Blabbermouth keeps it. Hold a shortcut, speak naturally, and your words arrive in the Windows field you were already using.

Join early access See it in motion

Quarterly update — NotesLocal

The revised launch plan gives us time to test the installer properly

Release to insert

This page moves when you do. Follow one sentence from microphone to finished text.

01 / Focus

Start where the words belong.

Click into an email, document, message or terminal. Blabbermouth remembers the destination.

Focused field found

02 / Speak

Hold. Talk. Stay in flow.

A compact listening overlay gives you enough feedback without becoming another application to manage.

Microphone active

03 / Local

The voice stays close.

Your chosen speech model processes the recording on this PC. There is no required cloud transcription trip.

Processing on device

04 / Release

Words land. You move on.

Release the shortcut. The finished transcription is inserted into the field you focused at the beginning.

Text inserted

Quarterly update — NotesField focused

Quarterly update — cursor ready

00:00 / 00:08

THE WHOLE WORKFLOW

Three moments. One natural gesture.

There is no blank page to open and no transcript to copy across. Your existing application remains the centre of the task.

01 / FOCUS

Place your cursor.

Choose the exact destination before you begin.

02 / SPEAK

Hold the shortcut.

Speak at your normal pace while the local overlay listens.

03 / RELEASE

Receive the text.

The transcription appears where your cursor was waiting.

THE PRIVATE ROUTE

No detour through someone else’s cloud.

Blabbermouth uses downloadable speech models on your Windows PC. The core journey is short, visible and yours.

01 / CAPTURE

Your microphone

Records only during the hold gesture.

02 / PROCESS

Your local model

Turns voice into text on this machine.

03 / INSERT

Your focused field

Receives the completed transcription.

No required accountThe everyday dictation workflow can work without a cloud login.

Optional local historyChoose whether recordings and transcripts are retained.

Visible model choicePick the balance of download size, speed and accuracy.

DESIGNED FOR EVERY DAY

The small things keep you talking.

Shortcuts, corrections, profiles and file transcription sit behind the simple hold-and-speak experience.

Activation

A shortcut that almost disappears.

Choose a comfortable hold gesture that does not interrupt typing.

CTRLALTHOLD

Corrections

Keep names and specialist terms close.

Add vocabulary you use frequently instead of correcting it every time.

Profiles

Change rules with context.

Save punctuation, formatting and model choices for different tasks.

Files

Bring longer recordings.

Transcribe supported audio and video files locally.

History

Return to recent words.

Search local transcript history when you choose to keep it.

Whisper
Small~466 MiB

Whisper
Medium~1.5 GiB

Whisper
Large v3~2.9 GiB

Parakeet TDT 0.6B v3
INT8~465 MiB

HONEST BOUNDARIES

Local does not mean limitless.

Performance depends on your hardware and selected model. Accuracy can vary with accents, specialist vocabulary and noisy recordings.

WINDOWS

Designed for Windows 10 and Windows 11.

SPEED

Faster hardware and smaller models return text sooner.

ACCURACY

Always review important names, numbers and specialist terms.

LANGUAGES

Coverage and quality vary by downloaded model.

Coming to Windows

Keep the thought. Lose the typing.

Join early accessPlaceholder contact · release date TBC