VoskSpeechRecognistion User Manual (UE5)

This manual is for Unreal Engine 5 users. It explains how to use the VoskSpeechRecognistion plugin from Blueprints to run speech recognition via a WebSocket-based Vosk server.

Preface (open-source dependency & what you must provide)

• This plugin leverages the Vosk ecosystem and protocol approach based on the open-source project alphacep/vosk-api.
• You must download and run a Vosk Language Server (or any compatible Vosk WebSocket server) yourself, and set ServerUrl to that server.
• You must download speech recognition models yourself from the official model page: Vosk Models, then configure the model path as required by your chosen Language Server.
• The default ServerUrl is ws://127.0.0.1:8080, matching the common default port for official Vosk speech-recognition WebSocket servers; change ServerUrl in FVoskRecognitionOptions if your server uses a different port.

Scope & prerequisites

• Engine: Unreal Engine 5.x
• Platform: Current implementation is primarily Windows-oriented (depends on AudioCapture and WebSockets)
• ASR server: You need a reachable Vosk WebSocket server (default: ws://127.0.0.1:8080)
  • The plugin includes an option for compatibility with vosk-language-server variants (see bSendConfigOnConnect below).

Install & enable

1. Place the plugin under your project Plugins/ directory (or the engine plugins directory).
2. In Unreal Editor, open Edit → Plugins and enable VoskSpeechRecognistion.
3. Restart the editor.

The plugin enables/depends on: