AI Dictation — Open-Source Voice-to-Text App

Native voice typing and speech-to-text for macOS, Windows, iPhone, iPad, and Android.

AI Dictation is an open-source voice-to-text app for macOS, Windows, iPhone, iPad, and Android. It turns speech into text in supported apps with offline speech recognition, optional cloud transcription and cleanup, personal vocabulary, writing rules, and spoken shortcuts.

Download for Mac or Windows · Install on iPhone or iPad · Install on Android · Watch the demo · Browse releases

Voice typing where you already work

AI Dictation provides native voice input controls for each platform:

• Use a configurable shortcut on macOS and Windows.
• Use the AI Dictation voice keyboard on iPhone and iPad.
• Keep your regular Android keyboard and add a floating microphone beside editable text fields.
• Dictate in supported messages, email, notes, documents, browsers, and other standard text fields.
• Choose offline recognition when supported or cloud transcription when you want it.
• Improve results with personal vocabulary, explicit replacements, writing rules, and spoken shortcuts.
• Apply optional cleanup and formatting after transcription.
• Keep local history for review, replay, and recovery where the platform supports it.

Platform behavior differs because macOS, Windows, iOS, and Android provide different ways to capture and insert text.

Download AI Dictation

Platform	Install	Requirements	Voice input
macOS	Download for Mac	macOS 13 or later	Configurable global shortcut
Windows	Download for Windows	64-bit Windows 10 or later	Configurable global shortcut
iPhone and iPad	Download on the App Store	iOS 15 or later	AI Dictation keyboard
Android	Get it on Google Play	Android 8.0 or later	Floating microphone

The iOS offline model and Live Activity require iOS 17 or later. Android APKs, Windows portable builds, checksums, and other release assets are available from GitHub Releases.

How AI Dictation works

1. Install the app and grant the permissions required by your platform.
2. Place the cursor in a supported text field.
3. Start recording with the desktop shortcut, iOS keyboard, or Android floating microphone.
4. Speak naturally and stop when you are finished.
5. AI Dictation transcribes the recording, applies any enabled cleanup and personal rules, and returns the finished text.

Offline speech recognition or cloud transcription

AI Dictation lets you choose how speech recognition runs on supported devices.

	Offline mode	Cloud mode
Speech recognition	Runs on the device	Sends the recording to the transcription service
Internet connection	Not required after the local model is ready	Required
Model setup	May require a one-time model download	No local speech model required
Optional cleanup	Cloud cleanup may still send the transcript when enabled	Cleanup may send the transcript for processing
Useful when	You need local recognition or have no connection	You choose cloud recognition and cleanup

Some platforms also provide Automatic mode, which chooses between available offline and cloud recognition.

Offline speech recognition and a fully offline workflow are not necessarily the same thing. If nothing should leave the device, confirm that cloud cleanup and other cloud writing features are disabled.

Personal vocabulary, replacements, and voice shortcuts

AI Dictation can use your reference context to improve the final transcript:

• Personal vocabulary supplies the intended spelling of names, products, acronyms, and specialist terms.
• Replacements map a spoken form to the text you want returned.
• Voice shortcuts expand a spoken trigger into a phrase you use frequently.
• Writing rules control formatting, tone, or structure for a particular context.

Reference terms are used only when the recorded speech supports them; adding a term does not tell the app to insert it into unrelated transcripts.

How AI Dictation differs from basic dictation software

Built-in dictation and browser speech recognition are useful for quick voice typing. AI Dictation is designed for people who want more control over where recognition runs and how the final text is shaped.

Need	AI Dictation approach
Dictate where you write	Platform-specific desktop, keyboard, and floating-microphone controls
Keep speech recognition local	Offline mode on supported devices
Clean up rough speech	Optional cleanup and formatting
Handle names and jargon	Personal vocabulary and explicit replacements
Reuse common phrases	Spoken shortcuts
Adapt output to context	Writing and app-context rules
Inspect or extend the implementation	Public source for the Apple, Windows, and Android apps

This comparison describes AI Dictation's implementation without making unsupported accuracy, speed, or “best app” claims.

Build from source

Clone the repository once:

git clone https://github.com/writingmate/aidictation.git
cd aidictation

Build for macOS or iOS

Requirements:

• macOS 13 or later
• Xcode 15 or later

cd Whishpermate
open Whispermate.xcodeproj

Select the macOS or iOS scheme in Xcode and run it.

Build for Windows

Requirements:

• Windows 10 or later
• Visual Studio 2022 with .NET desktop development tools, or the .NET 8 SDK

cd AIDictation.Windows
dotnet build AIDictation.sln

Build for Android

Requirements:

• Android Studio or Android SDK command-line tools
• JDK 17

cd AIDictationAndroid
./gradlew assembleDebug

Copy local.properties.template to local.properties only when a local build needs cloud-mode configuration. Do not commit API credentials.

Privacy and data processing

• Offline speech recognition processes audio on the device.
• Cloud transcription sends the recording to the transcription service.
• Optional cloud cleanup can send a transcript even when speech recognition ran offline.
• Some workflows keep recordings locally for history, replay, recovery, or retry; other temporary recordings are removed after successful processing.
• App credentials are stored using platform-provided secure storage.

Read the AI Dictation privacy policy for the complete and current terms.

Frequently asked questions

What is AI Dictation?

AI Dictation is a cross-platform voice-to-text app that lets you dictate into supported apps on macOS, Windows, iPhone, iPad, and Android. It combines offline or cloud speech recognition with optional cleanup, personal vocabulary, writing rules, and voice shortcuts.

Is AI Dictation open source?

Yes. The macOS, iOS, Windows, and Android source code is available in this repository under the MIT License. Bundled third-party components remain under their respective licenses.

Does AI Dictation work offline?

Offline speech recognition is available on supported devices and languages. A model download may be required first. Cloud cleanup can still send transcript text when enabled, so disable cloud features when the complete workflow must remain on-device.

Does AI Dictation work in every app?

It works in standard supported text fields, but operating-system and application restrictions apply. Password fields, secure fields, and apps that block accessibility or third-party keyboard input may not accept dictation.

Is AI Dictation a voice keyboard?

On iPhone and iPad, AI Dictation includes a voice-to-text keyboard. On Android, it adds a floating microphone while allowing you to keep your existing keyboard. The desktop apps use configurable shortcuts.

How is AI Dictation different from built-in speech recognition software?

AI Dictation adds selectable offline and cloud processing, optional cleanup, personal vocabulary, replacements, spoken shortcuts, context rules, local history, and a shared product across desktop and mobile platforms. No accuracy comparison is implied without a reproducible benchmark.

Is a free version available?

A free tier is available. Check aidictation.com for current usage limits and paid-plan details instead of relying on versioned pricing in this README.

Where can I report a problem?

Use GitHub Issues for reproducible bugs and feature requests. Do not include recordings, transcripts, credentials, or other sensitive data in a public issue.

Repository layout

AIDictationAndroid/      Android app
AIDictation.Windows/    Windows app
Whishpermate/            macOS and iOS apps
ci_scripts/              Shared release and signing helpers
docs/                    Architecture and reliability contracts
scripts/                 Build, release, and validation tools
references/              Release notes and supporting material

Contributing

Contributions are welcome. Read CONTRIBUTING.md before opening a pull request, and use SECURITY.md to report a vulnerability privately.

Project links

• AI Dictation website
• Download AI Dictation
• AI Dictation FAQ
• Privacy policy
• Release history
• Issue tracker
• YouTube channel

License

AI Dictation is available under the MIT License. See THIRD_PARTY_NOTICES.md for bundled components distributed under other licenses.