samuel-vitorino/sopro

A lightweight text-to-speech model with zero-shot voice cloning

What it solves

Sopro TTS is designed to provide high-quality, intelligible text-to-speech (TTS) and voice cloning with a very small footprint. It addresses the need for efficient, low-latency audio generation that can run on consumer hardware, such as laptop CPUs or directly in a web browser, without requiring massive GPU resources.

How it works

Sopro uses a lightweight 120M-parameter model (sopro-v2-turbo) that supports zero-shot voice cloning from a short reference audio clip (5-20 seconds). It can operate in two modes: an offline path for maximum quality and a streaming path for minimal latency, achieving a time-to-first-audio of approximately 300ms on a laptop CPU. The model supports English, European Portuguese, French, and German.

Who it’s for

This project is for developers building audio applications that require fast, on-device TTS and voice cloning, as well as those who need to integrate speech synthesis into browser-based environments via ONNX runtime.

Highlights

  • Small Footprint: 120M parameters, allowing it to run on laptop CPUs and in-browser.
  • Zero-Shot Cloning: Clones voices from 5-20 seconds of reference audio.
  • Multi-language Support: Supports English, European Portuguese, French, and German.
  • High Efficiency: Low Real-Time Factor (RTF) on M3 CPUs and H100 GPUs.
  • Streaming Capabilities: Supports streaming audio generation for real-time interactions.
  • Causal Vocoder: Includes a causal vocoder for the streaming path.

Related

  • Project
    tronghieuit/v-tts

    A lightweight Vietnamese text-to-speech system featuring multi-speaker synthesis and zero-shot voice cloning that runs efficiently on CPU.

  • Project
    OpenMOSS/MOSS-TTS-Nano

    A multilingual tiny speech generation model with 0.1B parameters designed for real-time, CPU-based inference and high-fidelity audio output.

  • Project
    Audio8-AI/Audio8_TTS

    A compact 0.6B-parameter multilingual text-to-speech model featuring zero-shot voice cloning and efficient deployment options for CPU and GPU.

  • Project
    RVC-Boss/GPT-SoVITS

    A powerful few-shot voice conversion and text-to-speech system that can clone a voice using as little as 5 seconds of audio.

  • Project
    KittenML/KittenTTS

    A lightweight, ONNX-based text-to-speech library that provides high-quality voice synthesis on CPU with models as small as 25MB.