Open Source Multilingual Text-to-Speech System

Github repositories

• collabora/WhisperSpeech

  WhisperSpeech: Open Source Multilingual Text-to-Speech SystemWhisperSpeech, formerly spear-tts-pytorch, is an open-source text-to-speech system based on Whisper by OpenAI. The model aims to be powerful and customizable, emphasizing stability for commercial use. Recent optimizations, including torch.compile integration, have boosted performance to over 12x real-time on consumer hardware. The project supports multilingual capabilities, demonstrated in a sample sentence blending English and Polish seamlessly. Additionally, there's an easy voice-cloning feature. You can test the system on Colab, and pre-trained models and datasets are available on HuggingFace.Link to Colab for testing