This is a UE5.1 project for demonstration of Whisper-based Real-time Speech Recognition, or WhisperRealtime
for short.
WhisperRealtime
is an Unreal Engine plugin for real-time speech-to-text transcription and alignment with multi-language support, based on OpenAI's Whisper model.
You can download packaged build of this demo project from here.
- Windows 10 64bit
- Unreal Engine 5.1.0
- WhisperRealtime plugin v1.0.0 or above
- Microphone connected to your PC
If you want to run with GPU,
- CUDA: 11.6
- cuDNN: 8.5.0.96
- Clone this repo:
git clone git@github.com:Akiya-Research-Institute/WhisperRealtime-Demo.git
- Open
WhisperRealtime-Demo/WhisperRealtimeDemo5.uproject
- Click
Content Drawer > Add > Add Feature or Content Pack...
- Select
Third Person
on Blueprint tab and clickAdd to Project
- Restart Unreal Editor.
- Click
Play
on Unreal Editor.
Select from the Windows (OS) setting.
Demo project contains 3 maps which corresponds to the 3 features described in "How to use" section of the manual.