LAPLACE Chat

This feature shipped on Dec 21, 2024

Features

• The most complete TTS integration in the universe, covering many languages, models, and voices, plus an open API any other provider can plug into free of charge — see General APIs on the Subspace Institute for details
• Queue mode and zero-latency mode, for streamers of every size

Queue Mode

The default read-aloud mode. Once on, new events are announced automatically, danmaku with identical content are merged dynamically according to queue size, and gifts, Super Chats, and guard events are read first (sorted by amount > time priority). The read-aloud queue holds at most 15 items, and on overflow danmaku events are dropped FIFO. Paid events such as gifts are exempt from the queue limit

Zero-Latency Mode

A LAPLACE Chat original that fully simulates real-time conversation and works with every external TTS service. Once on, all danmaku are read directly without passing through the queue, but several danmaku may end up being read at the same time, so it is not recommended for live rooms with heavy danmaku volume. Zero-latency mode also merges no danmaku, which raises the number of calls to the external TTS API — pick a service that fits your budget

Usage

Open the dashboard and turn it on from the settings panel in the sidebar

Performance Comparison

Below are voice samples from each TTS provider — press play to listen. To test mixed-language handling, here is the full test text used in this benchmark:

Benchmark updated: Apr 25, 2026

Custom API

If none of the TTS services integrated here meet your needs, you can implement TTS through a custom API. A custom API has to meet the following conditions:

• Accepts POST requests with a JSON body, and receives the token, text, voice, and instructions fields by default
• Returns the audio itself, in any audio format (.ogg, .mp3, etc.) your browser can play
• No restriction on the route — / or /v1/tts both work, whichever fits your project structure
• Serves over HTTPS, or sits on the same local network as the chat overlay and is reachable through localhost or 127.0.0.1
• Allows CORS requests from the chat overlay's address

Related Links

• fishaudio/s2-pro
• Qwen/Qwen3-TTS-12Hz-Base
• OpenMOSS-Team/MOSS-TTS-v1.5
• index-tts/index-tts
• bytedance/MegaTTS3
• nari-labs/dia
• FunAudioLLM/CosyVoice
• SparkAudio/Spark-TTS
• myshell-ai/MeloTTS
• synesthesiam/opentts
• Artrajz/vits-simple-api
• RVC-Boss/GPT-SoVITS
• coqui-ai/TTS
• sgl-project/sglang-omni