Focus a field
Click where the text should appear. Blabbermouth remembers the application and editable control.
Local speech-to-text for Windows 10 & 11
Ready in the system tray
Speak anywhere.Type everywhere.
Hold a shortcut, speak, then release. Blabbermouth transcribes on your PC and delivers the finished text into the field you were already using.
Version 1.9.2 · Windows 10/11 x64 · No account required
Project update
The release candidate passed its checks. I’ll share the installer this afternoon.
✓ Document field captured
Ready in the system trayLocal processingChoose your modelText history onlyNo cloud transcription
A natural part of Windows
From thought to field in three steps.
No special editor and no copy-and-paste loop. Your words return to the Windows field you chose once recording is complete.
Hold and speak
Use hold-to-talk, toggle mode, or a modifier-only shortcut. The overlay and tray icon make recording clear.
Release and continue
Your local model transcribes, applies enabled text rules, and delivers the completed result.
Local by design
Your voice takes the short route.
Normal dictation goes from your microphone, through a model on this PC, and back to your chosen field. There is no cloud transcription step.
Everything shown here happens on your PC
Microphone capture
Audio is recorded for the current dictation. Last-recording recovery audio stays in memory unless you choose Save WAV.
Local model processing
A downloaded Whisper or Parakeet model turns speech into text. The selected model stays prepared to reduce repeated-start latency.
Finished text delivery
Unicode keystrokes or a targeted native edit-control path returns text to the captured Windows field.
No accountSettings, models and statistics stay under your control.
No audio historyOptional history contains text transcripts, not microphone recordings.
Explicit persistenceHistory, exports and saved recovery audio are choices you make.
Dictate your way
Quiet controls. Useful depth.
Blabbermouth stays in the tray until you need it, with options that help the words come out the way you intended.
A trigger that never types into your terminal.
Use modifier-only activation such as Ctrl + Shift, or combine modifiers with Space, Tab, A–Z, or F1–F12. Keyed shortcuts suppress their primary key while active.
Ctrl+Shift
Always know when it’s listening.
A compact topmost overlay and pulsing red tray icon stay visible for the recording interval.
Make repeat misses deterministic.
Pair a heard phrase with the exact correction you want. Every replacement stays visible and editable.
“blabber mouth” → Blabbermouth
“power shell” → PowerShell
Expand the things you say every day.
Use exact triggers for reusable text, dates, times, line breaks and tabs.
“today stamp” → {{DATE}}
“signature block” → saved text
One voice. The right tone for each app.
Choose formal, casual, or verbatim output by destination process. Add optional spoken punctuation, filler removal, trailing spaces, and voice submission when each helps.
Email · FormalChat · CasualTerminal · VerbatimBrowser · Default
Models are optional downloads
Choose what runs locally.
Blabbermouth keeps its installer lean. See each model’s approximate storage before you download it after installation.
Whisper Small
The lighter Whisper option for everyday local dictation.
~466 MiBoptional downloadWhisper Medium
A larger Whisper option for users who favour accuracy.
~1.5 GiBoptional downloadWhisper Large v3
The largest included Whisper catalogue choice.
~2.9 GiBoptional downloadParakeet TDT 0.6B v3 INT8
A fast local alternative powered by sherpa-onnx.
~465 MiBoptional downloadWhisper supports optional translation into English and prefers supported NVIDIA CUDA with CPU fallback. Parakeet currently runs on CPU and transcribes without translation. Compatible ggml Whisper models can also be imported.
More than live dictation
Your text, ready to reuse.
Keep a small local history when it helps, or turn existing recordings and videos into transcripts and subtitles.
✓
Optional 50-item text history with search, favourites, editing, retype and Markdown or JSON export.
✓
Batch media transcription for WAV, MP3, M4A, AAC, WMA, MP4 and MOV.
✓
Non-destructive exports including transcript sidecars, SRT and WebVTT.
Blabbermouth — History & filesHistoryFile transcriptionExports
research-interview.m4aTranscript + SRTCOMPLETE
meeting-notes.mp3Transcript sidecarCOMPLETE
product-demo.mp4Whisper MediumPROCESSING
Clear operating boundaries
Built for real Windows fields.
Blabbermouth aims for broad usefulness without promising every control on every system.
Windows desktop only
Current support is Windows 10 and Windows 11 on x64 PCs.
Completed text after release
Live word-by-word typing is being explored, not shipped in version 1.9.2.
Matching integrity levels
An elevated target requires Blabbermouth to run elevated as well.
Compatibility varies by control
Designed for editable Windows fields, with no guarantee for every application or control.
Download a model first
Offline recognition begins after you install one of the optional local speech models.
Blabbermouth 1.9.2
Say it where you need it.
Install for Windows, choose your local model, then use one shortcut wherever you write.
[OFFICIAL DOWNLOAD URL REQUIRED] · Windows 10/11 x64