Ivan Anikin
55 lines
1.0 KiB
Markdown
55 lines
1.0 KiB
Markdown
# Qwen TTS Streaming
|
|
|
|
Runtime plugin for streaming Qwen TTS Gateway audio into Unreal Engine.
|
|
|
|
## Assets
|
|
|
|
Create two data assets in the editor:
|
|
|
|
- `Qwen TTS Streaming Settings Data Asset`
|
|
- `Qwen TTS Preset Data Asset`
|
|
|
|
They are registered in the Content Browser create menu under the Audio asset
|
|
category.
|
|
|
|
Default gateway URL:
|
|
|
|
```text
|
|
wss://services.ai-games-studio.ru/qwen-tts/generate/stream/ws
|
|
```
|
|
|
|
Default Kira preset id:
|
|
|
|
```text
|
|
23c7eff6-5a38-451c-9b31-2d312a4e8b3d
|
|
```
|
|
|
|
## Component
|
|
|
|
Add `Qwen TTS Streaming Component` to an actor. It inherits from `AudioComponent`.
|
|
|
|
Set:
|
|
|
|
- `Settings`
|
|
- `Preset`
|
|
|
|
Then call:
|
|
|
|
```text
|
|
Stream Text
|
|
```
|
|
|
|
The component opens a WebSocket, sends the request, receives JSON messages with
|
|
`audio_b64`, decodes each WAV chunk into PCM, and queues it into
|
|
`USoundWaveProcedural`.
|
|
|
|
Useful events:
|
|
|
|
- `On Stream Started`
|
|
- `On Audio Chunk`
|
|
- `On Stream Finished`
|
|
- `On Stream Error`
|
|
|
|
Each `On Audio Chunk` event includes the original `audio_b64`, decoded PCM byte
|
|
count, sample rate, channel count, and chunk index.
|