FluidVoice is a free open-source macOS dictation app that transcribes speech on the machine and types polished text into any app. Its repository describes it as a local alternative to Wispr Flow: one hotkey, local speech models, and an on-device cleanup pass that strips filler and matches tone per app.[1][2]
Quick verdict
Mac users who dictate daily and want it free, local, and cleaned up should choose FluidVoice: one hotkey types polished text into any app, Fluid-1 strips filler and matches per-app tone, and GPLv3 licensing means no subscription. The constraints are platform and scope: macOS 15 or later only with narrower model choice on Intel, single-user use with no team administration, and a privately maintained intelligence runtime. Privacy-led solo users and developers dictating into terminals fit best; Windows, mobile, or managed-team buyers should compare Wispr Flow or VoiceInk instead.[1][2]
Capabilities
System-wide hotkey dictation System-wide hotkey dictation: hold one hotkey and speak, and the result is typed into whichever app is focused, including terminals, code editors, mail, chat, and browsers.[1][2]
Local transcription models Local transcription models: Nemotron Speech 3.5, Parakeet, Apple Speech, and Whisper run on the machine, and the site states zero bytes leave the Mac during core dictation.[1][2]
Fluid-1 cleanup model Fluid-1 cleanup model: an optional custom-trained on-device model of about 3.5 GB removes filler, fixes casing and formatting, and adjusts tone, so rambling speech arrives as ready-to-send text.[1]
Three dictation modes Three dictation modes: Write Mode drafts prose, Command Mode drives voice control, and Direct Dictation types raw transcript, so writing and commanding stay in separate contexts.[1][2]
Per-app tone profiles Per-app tone profiles: output formatting and tone adapt to the active application, keeping messages, code comments, and documents styled differently without manual switching.[1]
Multilingual coverage Multilingual coverage: 40-plus languages across the supported model families, with exact coverage varying by model, so model choice decides language reach.[1]
Optional cloud providers Optional cloud providers: enhancement can route through user-supplied OpenAI, Groq, or custom provider keys, while offline core dictation stays fully local.[1]
Install and requirements Install and requirements: macOS 15 Sequoia or later on Apple Silicon or Intel, installed through Homebrew Cask, a GitHub release download, or a from-source build with Xcode.[1][2]
Who will love it
People dictating messages and drafts into any macOS app who want speech turned into polished text without switching tools.[1]
Privacy-led writers whose voice, audio, transcripts, and typed content must stay on the machine, with analytics limited to one anonymous daily signal plus optional totals.[1][2]
Developers dictating into terminals and code editors, the workflow shown in the official demo, who want hands-free drafting where they already work.[1]
Who should avoid it
macOS only: iPhone and Windows builds sit on a waitlist, so anyone outside the Mac gets no working app today.[1][2]
Hardware-gated models: Apple Silicon runs every model while Intel runs through Whisper, so older Macs get narrower model and language choice.[1][2]
Single-user scope: no team administration, shared workspace, enterprise plan, or SSO is documented, and support runs through Discord, issues, and discussions, so teams needing managed rollout should look elsewhere.[1][2]
Alternatives and compatibility
Wispr Flow is the vendor-endorsed comparison: the repository About text positions FluidVoice as a local Wispr Flow alternative with the same dictate-anywhere plus cleanup workflow kept on the Mac.[1][2]
VoiceInk is the closest editorial counterpart: both transcribe locally on the Mac, type into the focused app, remove filler, and apply per-app formatting, differing on cloud plan details that stay out of scope here.[1][2][3]
Switching is manual work across habits, hotkeys, vocabularies, and per-app profiles; no migration guide was observed, so budget setup time rather than expecting an import.[1][2]
Questions worth asking
What is the best FluidVoice alternative for non-technical users?
Of the documented relationships, VoiceInk is the closest Mac dictation counterpart with the same type-anywhere flow, and Wispr Flow is the incumbent FluidVoice positions against. Fully non-technical users should know FluidVoice still needs a Homebrew or download install plus microphone and accessibility permissions, so the zero-setup path is the dictation already built into macOS.[1][2][3]
Is FluidVoice really no-code?
How much does FluidVoice cost per month?
Zero dollars per month. FluidVoice is free forever with no tiers under GPLv3 for versions from 2026-02-23 onward, and continued work is funded through GitHub Sponsors. If you add your own OpenAI, Groq, or custom provider key for post-processing, that provider bills your own account, not FluidVoice. Checked October 9, 2026.[1][2]
Can you self-host FluidVoice?
That is its entire model, with no vendor cloud to sign into. The app runs on your own Mac with local transcription, needs no account, and installs through Homebrew Cask, a GitHub release download, or a from-source build. One honest boundary: the Fluid Intelligence runtime itself is privately maintained even though the app is GPLv3.[1][2][3]
Sources and editorial review
Source research. Reviewed on October 9, 2026. Written by Sanjib Sah. AI assisted; reviewed by a human editor.
We have not run or tested this product.
- FluidVoice, Free Open Source Voice-to-Text for macOS
official · Observed October 9, 2026
Official product site verified live 2026-10-09: free forever open-source macOS dictation, on-device transcription, Fluid-1 model, per-app tone, three modes, 40-plus languages, macOS 15 requirement, Homebrew install, free FAQ, offline statements, iOS and Windows waitlist, 100000-plus downloads.
- altic-dev/FluidVoice repository
official · Observed October 5, 2026
Official repository: About names the official site and positions the app as a local Wispr Flow alternative; README covers GPLv3 from 2026-02-23, features, models, Quick Start permissions and hotkey, requirements, source build, privacy with no voice or text collection, sponsors, and the disclaimed fluidvoice.org domain.
- FluidVoice LICENSE
official · Observed October 5, 2026
Repository license file contains the GNU General Public License version 3 text, matching the GPLv3 statement in the README and site.
- FluidVoice releases
official · Observed October 9, 2026
Release index observed 2026-10-09 with v1.6.10 as the latest visible release, evidencing active maintenance one patch ahead of the v1.6.9 intake record.
- FluidVoice Package.swift
official · Observed October 5, 2026
Swift package manifest declaring the FluidVoice package, macOS 15 platform, and Swift dependencies, evidencing a Swift codebase alongside the Xcode project.
- VoiceInk official site
official · Observed October 5, 2026
VoiceInk official site: local macOS dictation that writes into any app, removes filler, structures output, and applies per-app behavior, used as the counterpart workflow for the editorial competitor inference.