Voice Clone Studio

Canonical version: Voice Clone Studio.

Voice Clone Studio is a modular Gradio-based web UI for voice cloning, voice design, and multi-speaker conversation. It is powered by Qwen3-TTS, VibeVoice, and LuxTTS, and supports both Whisper and VibeVoice ASR for automatic transcription. Written in Python, licensed under Apache 2.0.

Features

• Voice Clone: Clone voices from short audio samples (3-10 seconds). Supports multiple engines (Qwen3-TTS 0.6B/1.7B, VibeVoice 1.5B/Large), voice prompt caching, seed control, 40+ emotion presets
• Conversation: Create multi-speaker dialogues. Qwen mode offers 9 preset voices; VibeVoice mode supports up to 4 custom speakers and up to 90 minutes of continuous speech. Useful for podcasts, audiobooks, and long-form narratives
• Voice Presets: Generate speech using pre-built premium voices with optional style instructions
• Voice Design: Create voices from natural language descriptions (age, gender, emotion, accent) without any reference audio
• Train Custom Voices: Fine-tune custom voice models with your own training data. Complete pipeline from dataset management to training and evaluation
• Prep Audio: Trim, normalize, convert to mono, denoise (via DeepFilterNet), and transcribe audio

Models

Models are automatically downloaded from HuggingFace on first use.

Requirements

• Python 3.10+
• CUDA-compatible GPU (8GB+ VRAM recommended)
• SOX and FFMPEG for audio processing
• Flash Attention 2 (optional, for faster generation)

References

• Source code: https://github.com/FranckyB/Voice-Clone-Studio
• Qwen3-TTS: https://github.com/QwenLM/Qwen3-TTS
• VibeVoice: https://github.com/microsoft/VibeVoice
• LuxTTS: https://github.com/ysharma3501/LuxTTS

Related

• Qwen3-TTS
• Voice Clone Lab — narrower alternative: one engine, one voice, dataset pipeline as the main event
• VibeVoice
• Text-to-Speech (TTS)
• Voice Cloning
• Gradio

About Sébastien

Ready to get to the next level?