No VaaniFlow account
Install Vaani and configure it without creating an account or signing in.
Vaani for Windows · Open-source dictation
Speak. Vaani writes where your cursor is.
Hold a shortcut, speak naturally, and release. Vaani transcribes into the Windows app you are already using, with optional cleanup, app-aware writing styles, dictionary terms, and reusable snippets.
Windows 10/11 · Your own Azure OpenAI Whisper deployment · Optional language model for cleanup and styles
The desktop workflow
Vaani stays out of the way until you speak.
Vaani is the Windows application. VaaniFlow is the open-source project behind it. The app listens on demand, processes the result, then returns you to the work already open.
Local app data
Settings, history, dictionary entries, snippets, insights, and profile details stay on this device.
Your provider
Speech and optional language-model requests use the Azure OpenAI deployments you configure.
Open source
The project is MIT licensed, developed in public, and available to inspect or contribute to.
Vaani has no hosted VaaniFlow account or sync service. Provider requests go directly to the Azure OpenAI resource you configure.
The default interaction
Hold. Speak. Release.
The sequence is intentionally short: stay in the application you are using and let the shortcut carry the text back to the cursor.
- 01Hold
Start at the cursor
Hold Ctrl + Win from the Windows text field where you want the words to appear.
- 02Speak
Say it naturally
Keep working in the same app while Vaani records and transcribes your thought.
- 03Release
Receive usable text
Release the shortcut. Vaani processes the result and inserts it at your cursor.
Prefer not to hold a shortcut? A quick tap starts hands-free recording; use the shortcut again when you are finished.
Built around the words
More than a microphone shortcut.
Vaani gives the transcript context after it arrives: how it should read, which words matter, what phrases should expand, and what happened over time.
Dictate without changing workspaces.
Use push-to-talk or hands-free recording while messages, documents, terminals, and forms remain in front of you.
- System-wide voice input
- Background chunking for longer recordings
- Terminal-friendly paste behavior
Keep the thought. Shape the delivery.
Optional cleanup can remove filler and adjust wording. App-aware styles help messages, email, and longer writing sound appropriate without changing the intent.
Requires the optional language-model deployment you configure.
Teach Vaani what you say often.
Keep names, jargon, acronyms, preferred spellings, and repeatable blocks of text close to the dictation workflow.
See what your voice produced.
Search local history and review word counts, dictation speed, streaks, and application usage without sending an analytics profile to VaaniFlow.
Inside Vaani
The product, exactly as it ships.
The desktop app keeps vocabulary, writing styles, reusable phrases, and everyday controls visible without turning dictation into another workspace to manage.
01 / Dictionary
Your language remains recognizable.
Save exact spellings and recurring terms, then let Vaani learn repeated names and acronyms as they appear in your dictation.
02 / Style
One thought, shaped for the app in front of you.
Choose how personal messages, work messages, email, and other writing should read after transcription.
03 / Snippets
Repeatable details stay ready on cue.
Speak a saved phrase naturally and insert its full expansion while the surrounding dictation keeps its polish.
04 / Settings
Set the behavior once. Keep speaking.
Shortcut, language, appearance, startup, sound, and notification controls stay local and explicit.
GeneralInput and language
AppearanceWindow and color
SystemStartup and sounds
Clear data boundaries
Local app data. Provider-routed speech.
VaaniFlow does not operate an account or sync service. Vaani stores its working data locally, while speech and optional cleanup use the Azure OpenAI resource you connect.
01MicrophoneRecorded on demand
audio→
02Your Whisper deploymentTranscribes the audio
transcript→
03VaaniProcesses locally
text→
04Current text fieldInserted at the cursor
Optional path: transcript → your configured language-model deployment → cleaned or styled text → Vaani
Stays on this device
- Settings and preferences
- Dictation history and insights
- Local profile details
- Dictionary entries
- Snippets
- Provider credentials
Sent when the feature runs
- Recorded audio goes to the configured Whisper deployment.
- Transcript text can go to the configured language-model deployment when cleanup or writing styles are enabled.
Azure availability, rate limits, and usage charges are governed by the Azure resource you connect. VaaniFlow does not proxy these requests or add a hosted subscription.
Before you install
Bring the speech provider. Vaani handles the desktop workflow.
Provider details are entered in Settings → Provider and saved locally. You can use dictation with Whisper alone; cleanup and writing styles are optional.
WindowsWindows 10 or 11
SpeechAn Azure OpenAI resource with a Whisper deployment
Optional polishA language-model deployment for cleanup and writing styles
Vaani is MIT licensed and has no VaaniFlow subscription. Usage charges from your configured Azure resource may apply.
Built in public
Inspect the path from microphone to cursor.
VaaniFlow is published under the MIT license. Read the implementation, follow releases, open an issue, or contribute an improvement.
Ready when the cursor is
Speak into the work already open.
Configure your Azure speech deployment, choose a shortcut, and start dictating into Windows text fields.