mic

Voxly

Download
Open Source

Dictate.
Format.
Paste.

Hold a hotkey, speak, release. Your words are transcribed, cleaned up by AI, and pasted into whatever app you're using. Free forever.

download Download Free star Star on GitHub
check Windows
check macOS
check Linux
Claude Code
1 $ Hold Ctrl+Space and speak:
2 "refactor the auth module to use jwt tokens instead of sessions and um make sure the middleware handles token refresh"
3 // Developer Mode output:
4 "Refactor the auth module to use JWT tokens
instead of sessions. Make sure the middleware
handles token refresh."
5 // Pasted into active app automatically
DEVELOPER MODE
Tauri v2
+
SolidJS
+
Rust
star Star
scale MIT LICENSE

Dictate in any context

Switch modes to format your voice for the task at hand.

mic yeah so basically i think we should probably refactor the auth middleware cause its kinda messy right now and maybe swap to using jwt tokens instead of sessions

I think we should refactor the auth middleware. It's messy right now, and we should swap to using JWT tokens instead of sessions.

model: llama-3.3-70b-versatile
You are a precise text editor that cleans up voice-dictated text. Remove filler words and verbal hesitations. Fix grammar, punctuation, and capitalization. Preserve the speaker's original wording, vocabulary, and tone. Do not rephrase or "improve" their language.
savings

No Subscription

Bring your own API key. Groq's free tier, OpenAI, or any OpenAI-compatible endpoint. You control the cost.

security

Private by Design

Dictation audio is never written to disk. Meeting recordings stay local until you click Transcribe. API keys live in your OS credential manager.

bolt

Lightweight

Built with Tauri and Rust. The entire app is around 10MB. No Electron, no bloat, launches instantly.

tune

Fully Customizable

Create custom modes with your own system prompts. Add vocabulary entries for words the model gets wrong.

New in v1.20

One hotkey. Full meeting capture.

Record screen, mic, and system audio as a local MP4. When done, click Transcribe — Voxly sends the audio to AssemblyAI and returns a speaker-labeled transcript with your voice and system audio on separate channels.

fiber_manual_record

Record

Press Ctrl+Alt+M. Screen, mic, and system audio recorded locally. No cloud at this step.

upload_file

Transcribe

Click Transcribe. Voxly extracts a dual-channel audio file — mic on channel 1, system audio on channel 2 — and uploads it to AssemblyAI for multichannel diarization.

record_voice_over

Speaker-labeled transcript

You and System — not generic Speaker A/B. Mic bleed from system audio is filtered automatically. Click any line to seek the video.

Format

MP4

Channels

2-ch AAC

Provider

AssemblyAI

Design Review — Jun 3, 2026

schedule48:22 hard_drive312 MB

Get started in seconds

Getting Started
1. Download the latest release from GitHub Releases
2. Run the app and open Settings
3. Add your API key — Groq (free) or OpenAI
4. Hold Ctrl+Space and speak
5. Release — clean text appears in your active app ✓

How Voxly compares

Feature
Voxly 		Wispr Flow Superwhisper
Pricing Free (BYOK) $15/mo $8.49/mo
Open Source check_circle cancel cancel
Custom Modes check_circle check_circle check_circle
Platforms Windows, macOS, Linux Mac, Windows Mac only
API Provider Any (BYOK) Built-in only Built-in only
Meeting Transcripts check_circle cancel cancel
Local Recording check_circle cancel cancel
bug_report arrow_forward

Report a Bug

Found something off? Open an issue on GitHub and help us improve.
devices arrow_forward

Test on macOS / Linux

Currently tested on Windows. We need testers on other platforms.
code arrow_forward

Contribute

MIT licensed. Fork it, improve it, submit a PR. Contributions welcome.