Voice & Speech Customization

• Built‑in 30+ multi‑style, multi‑gender and multi‑dialect voices to meet mainstream business scenarios such as customer service, broadcasting and reading.
• Large‑model‑powered TTS enables users to independently train personalized cloned voices by submitting audio samples for rapid voice customization.

Fine‑grained Speech Control

• Supports speech‑speed adjustment from level 0 to 100 (default: 50), enabling stepless fine control of broadcasting rhythm.
• The VITS model is compatible with SSML markup language, using tags such as phoneme, break and say‑as to precisely control pronunciation, pauses and specific readings.

Real‑time Speech Synthesis

• WebSocket streaming synthesis enables simultaneous synthesis and playback with millisecond‑level first‑packet response, supporting real‑time synthesis of both 600‑character short texts and ultra‑long texts.
• Large models intelligently predict text emotion and intonation, automatically matching corresponding emotional expressions to support high‑naturalness real‑time dialogue scenarios.

Enterprise‑level System Integration

• Offers multiple integration methods including HTTP single synthesis, WebSocket streaming synthesis, standard MRCP protocol and Java/C++ SDKs.
• Quickly connects with call center systems such as IVR and outbound calling via MRCP protocol, suitable for voice navigation, automatic outbound calls and content production scenarios.