Opus at 48kHz stereo, constant bitrate. The quality difference is insane
Just tested it — the noise suppression is so clean. Keyboard sounds are completely gone
That’s the Krisp integration right?
Yeah, it runs locally too. No audio gets sent to any third party
What about screen share? I noticed the preset selector got updated
6 presets from 480p30 up to lossless VP9 at 4K. The lossless mode does 20 Mbps
Voice that feels like the same room.
48kHz stereo Opus audio with Krisp AI noise suppression running locally on your device. Keyboard clatter, fans, and background chatter vanish — your voice stays untouched. Sub-45ms latency over LiveKit's globally distributed SFU.
Conversations that stay yours.
Every message, file, and reaction is encrypted end-to-end with AES-256-GCM before leaving your device. Rich text, emoji, reactions, and threaded replies — without compromising privacy.
just pushed the new voice engine. latency is down to 38ms
no way, that's insane. testing now
the noise suppression update is live too. keyboard sounds are completely gone
screen share looks crisp at 4K. huge improvement over last week
Shared listening. Soundtrack your team.
Listen together in real time. Queue tracks, skip, and control playback collectively — without leaving your voice channel. Everyone hears the same beat, in perfect sync.
Midnight City
M83
Do I Wanna Know?
Arctic Monkeys
Electric Feel
MGMT
Screen share without the compromise.
Share your screen at full fidelity — from 480p for low bandwidth to lossless VP9 at 4K 60fps pushing 20 Mbps. No Nitro paywall, no resolution caps. Every pixel, every frame.
Details
Every detail, considered.
Crystal Voice
48kHz stereo audio with constant bitrate encoding. Opus codec, no compromises.
End-to-End Encryption
ECDH P-256 key exchange with AES-256-GCM. Your conversations stay yours.
Lossless Screen Share
Up to 4K VP9 at 20 Mbps. Six quality presets from 480p to lossless.
Noise Suppression
Krisp-powered AI noise cancellation. Remove background noise without touching your voice.
Spatial Text
Rich messaging with inline emoji, file attachments, reactions, and threaded replies.
Zero Latency
LiveKit SFU architecture with WebRTC. Sub-100ms latency, globally distributed.
Performance
Voice chat that respects your machine.
Flux uses less memory at idle than Discord uses to show its loading screen. Smaller install, lower latency, higher audio quality — without draining your battery or hogging your CPU.
Memory Usage (Idle)(lower is better)
Memory Usage (Voice Call)(lower is better)
App Binary Size(lower is better)
Voice Latency (P95)(lower is better)
CPU Usage (Voice Call)(lower is better)
Audio Quality (Bitrate)
6.7×
Less memory
48 MB vs 320 MB idle
25×
Smaller install
12 MB vs 300 MB
2.7×
Lower latency
45ms vs 120ms voice P95
3.3×
Higher bitrate
320 kbps vs 96 kbps audio
Under the hood.
Flux is built on a Rust backend with LiveKit WebRTC for media routing. The desktop app uses Tauri — no Electron, no bloat.
Audio Codec
Opus
Sample Rate
48 kHz
Channels
Stereo
Bitrate Mode
CBR (constant)
Encryption
AES-256-GCM
Key Exchange
ECDH P-256
Media Transport
WebRTC (LiveKit SFU)
Video Codec
H.264 / VP9
Desktop Framework
Tauri (Rust)
Backend
Rust + SQLite
| Preset | Codec | Bitrate | FPS |
|---|---|---|---|
| 1080p60 | H.264 | 6 Mbps | 60 fps |
| 1080p30 | H.264 | 4 Mbps | 30 fps |
| 720p60 | H.264 | 4 Mbps | 60 fps |
| 720p30 | H.264 | 2.5 Mbps | 30 fps |
| 480p30 | H.264 | 1.5 Mbps | 30 fps |
| Lossless | VP9 | 20 Mbps | 60 fps |
Ready to switch? Join the voice chat that respects your privacy and your machine.
Available for macOS and Windows. Linux coming soon.