Optical Voice v0.1.0
The first working proof of concept for live speech transmitted through animated QR codes.
Optical Voice captures microphone audio, divides it into short recoverable groups, displays those groups as an animated optical stream, and reconstructs and plays the audio on another device while the sender is still speaking.
What works
- Live microphone capture
- Low-bitrate audio split into short independent groups
- Animated QR transmission
- Camera-based incremental decoding
- Luby Transform fountain-code recovery
- Integrity and sequence validation
- Buffered audio playback
- Stale speech is skipped instead of creating unlimited delay
- The same
/talk/page can transmit and receive simultaneously - Separate QR and camera panels with live Searching / Reading feedback
- Existing optical file and text transfer modes remain available
Demonstrated
Live optical audio was successfully transmitted from a phone screen to a laptop camera, reconstructed, decoded, and played while the sender was still speaking.
The application supports transmission and reception at the same time. Maintaining optical alignment in both directions remains experimental because both front cameras must continuously see the opposite screen.
No network carries the audio
The speech itself does not travel through:
- Wi-Fi
- Bluetooth
- Cellular service
- WebRTC
- A server
A network connection may only be used to initially load the web application.
Known limitations
- Simultaneous two-way alignment is physically difficult
- Latency is higher than a conventional phone call
- Browser audio-format compatibility varies
- Headphones are recommended to prevent feedback
- Performance depends on camera quality, autofocus, brightness, reflections, and device positioning
- The optical stream is not encrypted
Open source
Optical Voice is open source under the MIT License.
The project is based on Decimen Optical Transfer by Evan Crawley (Bash Alarmist). The original Git history, attribution, and MIT licence are retained.