RelayX

Try It Now

Story: The Birth of RelayX

I'm a loyal Tailscale user who frequently needs to remotely connect to high-performance PCs at home or in my dorm using portable devices. In my years of Tailscale experience, 99% of the time was spent using remote desktop. I deeply understand Tailscale's potential - once when I was in another city, I could still connect to my dorm PC with extremely low latency for remote desktop access. In various harsh network environments, I've witnessed Tailscale's excellent performance and reliability firsthand.
On the other hand, when gaming with friends, we were always troubled by the limitations of existing voice tools: poor audio quality, paid features, and privacy concerns... Tailscale seemed to perfectly solve these problems. Thus, RelayX was born. Its goal is simple: to create a voice tool that truly belongs to users and is not controlled by central servers.
Thanks to Tailscale's open-source strategy and the pion/webrtc implementation, I was able to turn this idea into reality. Although RelayX is currently just a very early project, I hope it can become an interesting and practical tool that demonstrates Tailscale's great potential in peer-to-peer communication.

Advantages: Why Choose RelayX?

• Efficient Client-side Neural Network Microphone Noise Reduction: Integrates RNNoise algorithm for real-time background noise elimination
• Unique Windows Process Audio Capture Feature: Can directly capture audio output from specific applications
• Zero Operating Cost: No central servers. The only bandwidth and performance limitations are your own network.
• Absolute Privacy: All voice data is end-to-end encrypted and transmitted directly between users, without going through any third parties.
• Excellent Connectivity: Thanks to Tailscale, RelayX can easily penetrate complex intranets and NAT environments, achieving "connect wherever there's internet".
• Complete Open Source: Trust comes from transparency. You can audit every line of code.

Architecture: How Does It Work?

RelayX's core is a very simple concept: combining Tailscale's network layer capabilities with WebRTC's communication abilities.

+---------------------------------------------+          
|  peer A                                     |          
|                                             |          
|                                             |          
|   +--------------------------------------+  |          
|   |  front end                           |  |          
|   +------------+-------------+-----------+  |          
|   | microphone |audio capture| screenshare  |          
|   +------------+----+------^-+-----------+  |          
|                     |      |                |          
|                 encoder  decoder            |          
|                     v      |                |          
|   +------------------------+---+---------+  |          
|   | go subprocess          |   | tailscale  |          
|   +------------------------+   | client  <--+----      
|   | pion/webrtc     <----------+         +--+--> peer B
|   +-----------------+----------+-^-+---+-^  |          
|                                  | |   | |  |          
+----------------------------------+-+---+-+--+          
                                   | |   | |             
                                   | |   | |             
                                   | v   v |             
                                 peer C  peer D          
            

Tech Stack Details

• Electron
• React
• shadcn/ui
• TailwindCSS
• Zustand

• RNNoise: Microphone noise suppression from xiph.org
• CPA: My Windows processes audio capture subprocess, check the repo for more info
• Pion/WebRTC: Go-based WebRTC library for handling P2P audio-video communication
• Tailscale: Embedded client providing secure virtual private network

Future Roadmap

History Dev Versions

This project has undergone several changes involving the core architecture.

• February 2025
• December 2025
• October 2024
• April 2024