Glow-TTS
4.6
Free Glow-TTS offers a fast and efficient solution for text-to-speech synthesis with good audio quality and controllable parameters.
A Generative Flow for Text-to-Speech via Monotonic Alignment Search, enabling fast, diverse, and controllable speech synthesis without external aligners.
0 views
0 saves
|
Refined 3/4/2026
About Glow-TTS
Glow-TTS is a flow-based generative model for parallel text-to-speech (TTS) that eliminates the need for external aligners. It leverages the properties of flows and dynamic programming to search for the most probable monotonic alignment between text and the latent representation of speech. This allows for robust TTS that generalizes to long utterances, and generative flows enable fast, diverse, and controllable speech synthesis. The model achieves an order-of-magnitude speed-up compared to autoregressive models like Tacotron 2. Glow-TTS can also be extended to multi-speaker settings. The implementation is based on PyTorch and includes configurations for training, inference, and integration with HiFi-GAN for improved vocoding and audio quality. The architecture incorporates monotonic alignment search and generative flows, ensuring efficient parallel processing and high-quality speech synthesis.
Core Capabilities
Glow-TTS is a flow-based generative model for parallel text-to-speech (TTS) that eliminates the need for external aligners.
Main Tasks
Flow-based Transformation
Explore all tools that specialize in flow-based transformation. This domain focus ensures Glow-TTS delivers optimized results for this specific requirement.