Speech-AI-Forge

Why we included this project

Running more than one speech model usually means juggling separate installs and APIs. Speech-AI-Forge pulls several TTS engines, including CosyVoice, Fish-Speech, Index-TTS, Spark-TTS, and GPT-SoVITS, behind a single API server and a Gradio web UI, so you can swap models without reworking your pipeline. It also includes speech recognition via SenseVoice and Whisper, which makes it handy for voice-in, voice-out experiments rather than plain one-way generation. Deployment is flexible: a one-click Colab notebook, a Windows bundle, and a Docker image all get you running without wrestling with environment setup. For teams prototyping voice assistants or audio workflows, the unified server interface is a convenient way to drive several engines from one codebase.