VOX Screen
Realtime Speech-to-Screen.

Display live captions and translation on-screen in realtime with offline AI powered software.

AI generated realtime Speech-to-Text captions and translation content display software solution.

Designed by events professionals for live captions and translation display for live events, education and government use.

High-performance realtime “Speech to Text” (STT) App
build on OpenAI’s automatic speech recognition (ASR) models.

Watch VOX Screen DEMO

Minimum system requirements: Intel i7, 16G RAM, NVIDIA GeForce GTX1060 Ti – Model used: Small
NVIDIA GeForce RTX 4060Ti – Model used: Medium
For large models high spec GPU recommended.

NOTE:
Install CUDA for better performance (windows only)

Tested with: Apple Mac Book Pro M2 and M3 chips.
Total Number of Cores: 12
(8 performance and 4 efficiency)
Memory: 16 GB
Total Number of GPU Cores: 19 (Metal 3)
Models in use: All models

1 Year license (365 days) £250

Free upgrades
and customer support
for 1 year.

(Activation key will be e-mailed upon completion of the order)

Lifetime License £350

Free updates on current version fixes and improvements.
50% off for the next generation paid upgrades and full customer support

(Activation key will be e-mailed upon completion of the order)

// Main Features:

Customisable “Program out” output via HDMI/Display port
Compliant (data processed locally)

Offline Transcription (no internet required)
Low latency realtime captions / speech-to-text generation
Customisable program output: text positioning, font size + colour, background colour
High accuracy / Low word error rate with Medium and Large models
Build on High performance OpenAI’s automatic speech recognition (ASR) model
Record transcription to file

//Things we are working on:

Removing some “hallucinations” on blank audio
Replacing audio meeter with more accurate audio -dB UI
Building in direct incoming audio monitoring via headphones output or system audio output.

// Windows

Efficient GPU utilisation for NVIDIA cards
CUDA accelerated model processing
CPU Only processing

Minimum system requirements: Intel i7, 16G RAM, NVIDIA GeForce GTX1060 Ti (or similar) – Model used: Small

NVIDIA GeForce RTX 4060Ti – Model used: Medium
For large models high spec GPU recommended.

NOTE: Install CUDA for better performance

// MacOS

Efficient GPU utilisation on Apple Silicon M1, M2 and M3 chips
METAL 3 accelerated model processing

Tested with: Apple Mac Book Pro M2 and M3
Number of Cores: 12
(8 performance and 4 efficiency)
Memory: 16 GB
Total Number of GPU Cores: 19 (Metal 3)
Models in use: All models

VOX ScreenRealtime Speech-to-Screen.