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

AI Dictation is an MIT-licensed, open-source speech-to-text app for macOS, Windows, iPhone, iPad, and Android. It provides voice typing in supported apps with offline recognition, optional cloud transcription and cleanup, personal vocabulary, writing rules, and spoken shortcuts. Feature availability varies by platform and workflow.

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

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.
• Use the personal-vocabulary, replacement, writing-rule, and spoken-shortcut controls available on your platform.
• 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.

The macOS app installs on macOS 13 or later, while offline recognition requires macOS 14 or later. 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.

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.

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

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.

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.

AI Dictation instructs cleanup to use reference terms only when the recorded speech supports them; adding a term is not intended to insert it into unrelated transcripts.

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.

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

Clone the repository once:

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

Requirements:

• Xcode 26 or the current Xcode version used by the Apple CI workflow
• A macOS version supported by that Xcode release

cd Whishpermate
open Whispermate.xcodeproj

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

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

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.

• 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.
• Account session tokens and user-entered API keys are stored using platform secure storage where applicable.

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

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

Yes. AI Dictation's native client source is available under the MIT License: macOS and iOS, Windows, and Android. Bundled third-party components remain under their own licenses; see Third-Party Notices.

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.

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.

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.

Depending on the platform and workflow, AI Dictation can add selectable offline or cloud processing, optional cleanup, personal vocabulary, replacements, spoken shortcuts, context rules, and local history. No accuracy comparison is implied without a reproducible benchmark.

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.

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

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

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

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

AI Dictation's native client source is available under the MIT License. See THIRD_PARTY_NOTICES.md for bundled components distributed under other licenses.