Real-Time Voice Cloning

Clone a voice from five seconds of audio and synthesize arbitrary speech in real time.

Real-Time Voice Cloning is a widely studied implementation of the SV2TTS framework, cloning a voice from roughly five seconds of reference audio and generating new speech with it. It chains a speaker encoder, a Tacotron synthesizer, and a vocoder, and remains a popular reference for understanding the encoder-synthesizer-vocoder TTS architecture.