Shypd.aiThe largest open index of AI toolsGLM-TTS is an open-source text-to-speech (TTS) synthesis system that uses large language models to generate controllable and emotion-expressive speech. It supports zero-shot voice cloning and streaming inference for real-time applications.
Trending
About
GLM-TTS is a high-quality text-to-speech (TTS) synthesis system built on large language models, offering zero-shot voice cloning and streaming inference capabilities. Its two-stage architecture first employs an LLM to create speech token sequences, then uses a Flow model to convert these into high-quality audio waveforms. A key differentiator is its Multi-Reward Reinforcement Learning framework, which significantly enhances emotional expression and prosody, moving beyond the flat delivery of traditional TTS systems. It supports real-time audio generation, making it suitable for interactive applications, and offers multi-language support, primarily Chinese with English mixed text. The system also features phoneme-level modeling for fine-grained pronunciation control, addressing ambiguities in polyphones and rare characters through a 'Hybrid Phoneme + Text' input mechanism.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
