# FluidVoice

FluidVoice is a fully open source voice-to-text dictation app for macOS with local speech models, AI post-processing via the Fluid-1 model (Fluid Intelligence), adaptive tone per app, global hotkeys, smart typing into any app, command mode, write mode, live preview, history, and stats.

Canonical URL: https://altic.dev/fluid
Markdown URL: https://altic.dev/fluid.md
Source-of-truth README: https://github.com/altic-dev/FluidVoice
Latest release: https://github.com/altic-dev/Fluid-oss/releases/latest
Build/source repository from README: https://github.com/altic-dev/Fluid-oss

## Quick Facts

- Platform: macOS 15.0 Sequoia or later
- Hardware: Apple Silicon; Intel supported from 1.5.1 builds using Whisper
- Price: Free
- License: GPLv3 from 2026-02-23 onward
- Primary models: Nemotron Speech 3.5, Parakeet Flash, Parakeet TDT v3/v2, Cohere Transcribe, Apple Speech, Whisper
- AI enhancement: Fluid Intelligence with Fluid-1 model (local, on-device) or optional cloud providers (OpenAI, Groq, custom)
- Privacy: Local-first; optional anonymous analytics; optional AI provider keys stored in macOS Keychain

## Requirements

- macOS 15.0 (Sequoia) or later
- Apple Silicon Mac (M1, M2, M3, M4)
- Intel Macs supported from 1.5.1 builds using Whisper models
- ~1 GB disk space for a voice model
- ~3.5 GB disk space for the Fluid Intelligence model (Fluid-1, optional)
- Microphone access
- Accessibility permissions for typing

## Fluid Intelligence and the Fluid-1 Model

Fluid Intelligence is powered by the Fluid-1 model — an on-device AI post-processing model for dictation.

Fluid Intelligence is a separate, privately maintained local AI runtime powered by the Fluid-1 model. It delivers advanced on-device dictation enhancement — smart formatting, context-aware capitalization, adaptive tone based on the active app, and post-processing — all running locally on your Mac.

### Key capabilities

- Fluid-1 model — on-device AI post-processing for smart formatting, context-aware capitalization, and adaptive tone
- Detects the active app and rewrites dictation to match the expected tone (casual for Slack, formal for Mail, structured for GitHub, etc.)
- No cloud, no API keys, no data leaving your Mac
- Fully optional — the core dictation experience works without it
- ~3.5 GB download, runs entirely on-device

- Disk size: ~3.5 GB
- Privacy: Fluid Intelligence (Fluid-1) runs fully locally. Zero voice data, text, or prompts leave your Mac.
- Licensing: FluidVoice is GPLv3 open source. Fluid Intelligence and the Fluid-1 model are a separate privately maintained runtime, kept private for now to sustainably offer the core dictation experience for free.

## Features

- Fluid Intelligence with Fluid-1 model — on-device AI post-processing for smart formatting, context-aware capitalization, adaptive tone per app, and post-processing, all running locally with zero data leaving your Mac
- Adaptive tone — Fluid-1 detects the active app and rewrites dictation to match the expected tone (casual for chat, formal for email, structured for code reviews)
- Live Preview Mode with real-time transcription preview in the overlay
- Multiple speech models: Nemotron Speech 3.5, Parakeet Flash, Parakeet TDT v3 and v2, Cohere Transcribe, Apple Speech, and Whisper
- Real-time transcription with extremely low latency
- AI enhancement with Fluid Intelligence / Fluid-1 (local), OpenAI, Groq, and custom providers
- Global hotkey for instant voice capture
- Smart typing directly into any app
- Menu bar integration
- Auto-updates with seamless restart
- Opt-in beta channel
- Command Mode for taking actions on the Mac by voice
- Write Mode for writing or rewriting text in any text box
- History stats and usage stats

## Supported Speech Models

| Model |Best for |Language support |Download size |Hardware |
| --- | --- | --- | --- | --- |
| Nemotron Speech 3.5 — Ultra Fast Low Latency | Streaming-capable multilingual dictation | ~40 languages | ~670 MB | Apple Silicon |
| Nemotron 3.5 Multilingual | Higher-accuracy multilingual dictation | ~40 languages | ~530 MB | Apple Silicon |
| Parakeet Flash (Beta) | Lowest-latency live English dictation | English only | ~250 MB | Apple Silicon |
| Parakeet TDT v3 | Fast default multilingual dictation | 25 languages | ~500 MB | Apple Silicon |
| Parakeet TDT v2 | Fastest English-only dictation | English only | ~500 MB | Apple Silicon |
| Cohere Transcribe | High-accuracy multilingual dictation | 14 languages | ~1.4 GB | Apple Silicon |
| Apple Speech | Zero-download native macOS speech recognition | System languages available on the user's Mac | Built in | Apple Silicon and Intel |
| Whisper Tiny / Base / Small / Medium / Large | Broad compatibility, including Intel Macs | Up to 99 languages depending on selected model size | ~75 MB to ~2.9 GB | Apple Silicon and Intel |

## Model Notes

- Nemotron Speech 3.5 is the top pick for streaming-capable, multilingual dictation with ~40 language support.
- Parakeet Flash is the best pick when words need to appear live with the lowest latency.
- Parakeet TDT v3 remains the default multilingual Apple Silicon model.
- Cohere Transcribe is the stronger pick for a larger multilingual model with higher displayed accuracy.
- Whisper remains the fallback for Intel Macs and the widest language coverage.

## Language Details

### Nemotron Speech 3.5 Languages

Approximately 40 languages including English, French, German, Italian, Spanish, Portuguese, Mandarin, Japanese, Korean, and more.

### Parakeet Flash Languages

English.

### Parakeet TDT v3 Languages

Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, and Ukrainian.

### Parakeet TDT v2 Languages

English.

### Cohere Transcribe Languages

English, French, German, Italian, Spanish, Portuguese, Greek, Dutch, Polish, Mandarin, Japanese, Korean, Vietnamese, and Arabic.

### Apple Speech Languages

System language support depends on the macOS speech recognition languages available on the user's machine.

### Whisper Language Support

Whisper supports up to 99 languages, depending on the selected model size.

## Quick Start

1. Download the latest release.
2. Move FluidVoice to the Applications folder.
3. Grant microphone and accessibility permissions when prompted.
4. Set a preferred hotkey in settings.
5. Optionally add an AI provider API key for enhanced transcription. Keys are stored securely in macOS Keychain.
6. Optionally opt in to beta builds in Settings -> Automatic Updates -> Beta Releases.

## Privacy and Analytics

FluidVoice is local-first. Optional anonymous analytics exist for usage, reliability, and release health, but the README states that voice, raw audio, transcribed text, selected text, prompts, AI-generated responses, terminal commands, window titles, file paths, clipboard content, and typed content are not collected.

### Anonymous analytics may collect

- App version, build, and macOS version
- Low-cardinality feature/config flags, such as app mode and major settings
- Approximate usage ranges
- High-level success/error outcomes

### Anonymous analytics does not collect

- Voice, raw audio, or transcribed text
- Selected text, prompts, or AI-generated responses
- Terminal commands or outputs
- Window titles, file names, file paths, clipboard content, or typed content
- Personal or private information

Disable path: Settings -> Share Anonymous Analytics

## Demo Transcript Summaries

- Command Mode: Command Mode lets a user speak an instruction and have FluidVoice take an action on the Mac.
- Write Mode: Write Mode lets a user write or rewrite selected text in any text box in any Mac app.
- Dictation Mode: Dictation Mode captures speech and types the resulting text into the active text field.

## FAQ

### Is FluidVoice free?

Yes. FluidVoice is completely free and open source.

### What macOS version does FluidVoice require?

FluidVoice requires macOS 15.0 Sequoia or later, plus microphone access and accessibility permissions for typing.

### Does FluidVoice support Intel Macs?

Yes. Current version supports Intel Macs by using Whisper models.

### Which speech models does FluidVoice support?

FluidVoice supports Nemotron Speech 3.5 (Ultra Fast Low Latency and Multilingual), Parakeet Flash (Beta), Parakeet TDT v3, Parakeet TDT v2, Cohere Transcribe, Apple Speech, and Whisper Tiny/Base/Small/Medium/Large.

### What is Fluid Intelligence and the Fluid-1 model?

Fluid Intelligence is a local AI runtime which runs the Fluid-1 model. It delivers advanced on-device dictation enhancement — smart formatting, context-aware capitalization, adaptive tone based on the active app, and post-processing — all running locally on your Mac. It is fully optional (~3.5 GB download), requires no cloud or API keys, and no data leaves your machine. FluidVoice works great on its own without it.

### How many languages does FluidVoice support?

Language support is model-specific: Nemotron Speech 3.5 supports ~40 languages, Parakeet Flash and Parakeet TDT v2 are English-only, Parakeet TDT v3 supports 25 languages, Cohere Transcribe supports 14 languages, Apple Speech depends on macOS system languages, and Whisper supports up to 99 languages depending on model size.

### Does FluidVoice work offline?

FluidVoice is local-first and uses local speech models for dictation. AI enhancement can use optional providers such as OpenAI, Groq, or custom providers, so cloud use depends on the features and providers a user enables.

### What permissions does FluidVoice need?

FluidVoice needs microphone access for voice capture and accessibility permissions for typing into other apps.

### What license does FluidVoice use?

From 2026-02-23 onward, FluidVoice is licensed under the GNU General Public License v3.0 (GPLv3). Versions before that date used Apache License 2.0.

### What does FluidVoice analytics collect?

Optional anonymous analytics, FluidVoice does not collect voice, raw audio, transcribed text, prompts etc. No sensitive info leaves your mac

### Where can I download FluidVoice?

You can click the download button or get the latest release at https://github.com/altic-dev/Fluid-oss/releases/latest.