Connect runs quietly in the background and translates your voice in real time — no complex setup, no technical knowledge required.
No audio routing needed. Connect works directly with your system's audio pipeline.
Your voice is translated and routed to your call in real time. Others hear your translated voice.
Hear the translated audio yourself. Useful to understand what others are saying in your language.
Download and install the Connect app for Linux. It integrates directly with PipeWire/PulseAudio — no extra tools needed.
Zoom, Meet, Discord, Teams — whatever you use. Just open it and join or start a call normally.
In Connect, press the big CONNECT-ON button. That's it. Your voice is now being translated live for everyone on the call.
Use Connect to translate your voice for others, or monitor translated audio yourself.
Others hear your translated voice in the call. Set Connect as your mic & speaker in your app or system.
Hear the translated audio yourself only. Use "How I Sound?" to test your voice before a call.
Download and run the Connect installer. It registers virtual audio devices automatically.
Option A — In your call app: Go to audio settings and select Connect
Microphone and Connect Speaker.
Option B — System default: Go to Windows Sound Settings → set Connect
as default input and output. Every app uses it automatically from then on.
Physical mic
Real-time AI
Hears translation
Press CONNECT-ON and speak naturally. When you disconnect, your system reverts to your original devices automatically — nothing to undo.
Use Connect to translate your voice for others, or monitor translated audio yourself.
Others hear your translated voice. Set Connect as your audio device in your call app or system settings.
Hear the translation yourself only. Great for testing your voice or understanding incoming speech.
Download the macOS app. It installs a virtual audio driver compatible with all macOS versions.
In-app: Open your call app's audio preferences and select Connect
Microphone / Connect Speaker.
System default: System Settings → Sound → set Connect as input and
output. Every app uses Connect automatically from then on.
Built-in or external mic
Real-time AI
Hears your voice
Hit CONNECT-ON and speak. When you disconnect, your Mac switches back to your original audio devices automatically.
Optional settings
Everything works out of the box. These settings let you go further when you need it.
Choose which voice others hear — your cloned voiceprint, a custom synthetic, or the default. Swap anytime.
One-way (you speak, they hear translated) or two-way (both sides translate in real time).
Connect learns from the conversation to improve translation accuracy over time during your session.
Tell Connect your domain — medical, legal, tech — for more accurate terminology translation.
Automatically label who's speaking in your transcription — names, roles, or "Speaker A / B".
Preserve your tone, energy, and expressiveness across language — your enthusiasm comes through.
Voice Identity
Connect preserves who you sound like across translations. Three ways to set up your voice — from simple to fully custom.
Clone your actual voice
Upload a 10-second audio sample or record yourself directly. Connect clones your voice so translations sound authentically like you — your tone, timbre, and character preserved across every language.
Remix an existing voice
Take any voice from your library and transform it — add a different accent, make it deeper, warmer, more energetic. Age and gender stay anchored to the original, only the delivery style changes.
Create from scratch
Design a completely synthetic voice by defining labels (age, gender, accent, language) and describing its personality. Connect generates 3 unique voices matching your spec — choose your favourite.
Fine-tune how Connect listens and delivers your translation.
Tell Connect which languages you expect to hear during the call. This helps the AI prioritize them and reduces misdetection — especially useful in multilingual meetings or when speakers switch languages frequently.
Waits until you finish speaking, then sends the full sentence for translation. More accurate, slightly delayed.
Sends audio continuously as you speak for ultra-low latency. Best for live conversations where speed matters.