FluidVoice FAQ

FluidVoice can be installed in two ways:

1. Using Homebrew: Run brew install --cask fluidvoice
2. Manual download: Visit the GitHub releases page and download the latest version

After installation, grant microphone and accessibility permissions when prompted.

FluidVoice requires:

• macOS 15.0 (Sequoia) or later
• Apple Silicon Mac for all speech models
• Intel Macs supported via Whisper models (from 1.5.1+)
• ~1 GB disk space for voice models
• ~3.5 GB disk space for Fluid Intelligence model (optional)
• Microphone access
• Accessibility permissions for typing

After installation and permissions:

1. Open FluidVoice
2. Go to Settings and set your global hotkey
3. Complete the onboarding flow to select your speech model
4. Optionally enable Fluid Intelligence for on-device AI enhancement
5. Configure your preferred settings

Press your hotkey and start dictating!

Fluid Intelligence is a proprietary local AI runtime that powers advanced on-device dictation enhancement including:

• Smart formatting (quotes, dashes, ellipses)
• Context-aware capitalization
• Post-processing for cleaner transcripts
• Local processing with zero data leaving your Mac

It's completely optional and works alongside any supported speech model.

Command Mode: Lets you control your Mac by voice - launch apps, run shortcuts, trigger system actions, and automate workflows without touching the keyboard.

Write Mode: Lets you write or rewrite text directly in any text field across any app. Select text and rewrite it inline or dictate new content.

Both modes can be activated via the global hotkey or from the menu bar.

Yes! FluidVoice is local-first by default:

• All speech-to-text processing happens on-device
• Fluid Intelligence runs entirely locally
• No data leaves your machine unless you explicitly opt in to cloud AI providers

You can use it completely offline without any internet connection.

Yes! While most speech models require Apple Silicon for optimal performance, Intel Macs are supported via:

• Apple Speech (built-in)
• Whisper models (Tiny, Base, Small, Medium, Large)

Intel Macs can still use all features except Fluid Intelligence and certain speech models.

The best model depends on your needs:

• Parakeet Flash: Lowest latency for English dictation (~250MB)
• Nemotron Speech 3.5: Fastest multilingual streaming (~670MB)
• Parakeet TDT v3: Good balance of speed and multilingual support (~500MB)
• Apple Speech: No download required, uses system languages
• Whisper: Broadest language support (99 languages), Intel-compatible

Try different models during onboarding to see which works best for you.

Currently, FluidVoice supports the pre-configured models mentioned in our documentation. While we don't support arbitrary custom models, we're always evaluating new models and adding support based on user requests and technical feasibility.

If you have a specific model you'd like to see supported, please create an issue on GitHub.

Speech models vary in size:

• Apple Speech: Built-in (no download)
• Whisper Tiny: ~75MB
• Parakeet Flash: ~250MB
• Parakeet TDT v3/v2: ~500MB
• Nemotron Speech 3.5: ~670MB
• Cohere Transcribe: ~1.4GB
• Whisper Large: ~2.9GB

Fluid Intelligence requires ~3.5GB if enabled.

By default, all voice processing happens on your Mac:

• Speech-to-text: On-device using local models
• Fluid Intelligence: On-device AI enhancement
• Text processing: On-device formatting

Only if you explicitly enable cloud AI providers (OpenAI, Groq, etc.) will your data be sent to external services. All API keys are stored securely in macOS Keychain.

FluidVoice collects minimal anonymous data to track app health:

• App version and build number
• macOS version
• General feature usage (low-cardinality)
• Approximate usage ranges
• High-level success/error outcomes

Never collected: Voice data, raw audio, transcribed text, selected text, prompts, AI responses, terminal commands, window titles, file paths, clipboard content.

You can disable analytics at any time in Settings.

Yes, FluidVoice is designed with privacy in mind:

• No personal data processing by default
• Opt-in analytics with clear controls
• Local-first processing architecture
• No user tracking or profiling

The app follows privacy-by-design principles and doesn't collect identifiable personal information.

Try these troubleshooting steps:

1. Check microphone permissions in System Settings
2. Verify accessibility permissions for FluidVoice
3. Restart the FluidVoice app
4. Try a different speech model
5. Check for app updates
6. Review logs in Console.app (search for "FluidVoice")

If issues persist, create an issue on GitHub with your system info and error details.

Bugs and feature requests should be reported via GitHub:

• Visit Issues page
• Create a new issue with detailed description
• Include macOS version, app version, and steps to reproduce
• For bugs, include console logs if possible

Join our Discord community for real-time support and discussions.

Currently supported: macOS 15.0+

In development: iOS, Windows, and Linux versions are planned but not yet released.

Follow development updates on X (@ALTIC_DEV) or GitHub for platform availability announcements.

Contributions are welcome! Here's how to help:

• Star the repository on GitHub
• Report bugs or suggest features
• Submit pull requests for bug fixes or enhancements
• Help with documentation improvements
• Translate the app into other languages
• Support development via GitHub Sponsors

Check our GitHub page for contribution guidelines and development setup instructions.

FluidVoice is licensed under GNU General Public License v3.0 (GPLv3) from February 23, 2026 onward.

Versions published before this date were licensed under Apache License 2.0.

The license is included in the repository and ensures the software remains free and open source.

To build FluidVoice from source:

1. Clone the repository: git clone https://github.com/altic-dev/FluidVoice.git
2. Navigate to the project: cd FluidVoice
3. Open the project: open Fluid.xcodeproj
4. Build and run in Xcode

All dependencies are managed via Swift Package Manager. For unsigned builds, use: xcodebuild -project Fluid.xcodeproj -scheme Fluid -destination 'platform=macOS' build CODE_SIGNING_ALLOWED=NO