Favicon of Voice-Clone-Studio

Voice-Clone-Studio

A browser-based AI voice cloning studio combines Qwen3-TTS, VibeVoice and other speech engines with voice design, conversion and custom voice training.

Alternative to:

Voice-Clone-Studio brings voice cloning and speech production into one browser interface for people making podcasts, audiobooks and dialogue recordings. It combines several engines with tools for preparing samples and training custom voices.

  • Voice cloning and design: Generate speech from reference recordings with Qwen3-TTS, VibeVoice, LuxTTS, Chatterbox or Fish Speech S2 Pro, or create a voice from a written description with Qwen3-TTS.
  • Dialogue and delivery: Build scripts with preset or custom speakers, adjust emotion, and use Fish Speech expression tags for delivery such as whispering or laughter; VibeVoice supports long recordings with custom voices.
  • Voice conversion: Chatterbox changes the voice in existing speech using a target recording, with an English-focused model and a multilingual alternative.
  • Custom voice training: Fine-tune Qwen3-TTS or VibeVoice using recordings and transcripts, then generate speech with the trained voices alongside presets.
  • Audio preparation: Trim and normalize recordings, convert them to mono, remove noise with DeepFilterNet and extract audio tracks from video.

Automatic transcription works with Qwen3-ASR, VibeVoice-ASR and Whisper. The studio also includes MMAudio for sound effects and llama.cpp and Ollama for prompt generation. Individual tools can be enabled or disabled within the interface.

Voice-Clone-Studio is open source under Apache 2.0. Custom voice training requires a CUDA GPU and multiple audio samples with transcripts. VibeVoice dialogue generation may add background music or sounds without an explicit request.

Similar to Voice-Clone-Studio