explaingit

openbmb/voxcpm

📈 Trending★ 19,144PythonAudience · developerComplexity · 2/5ActiveLicenseSetup · moderate

TLDR

Text-to-speech system that generates natural, expressive speech directly from text without tokenization, supporting voice design, cloning, and 30 languages.

Mindmap

mindmap
  root((VoxCPM))
    What it does
      Text to speech
      Voice design
      Voice cloning
    Key features
      Diffusion models
      Autoregressive generation
      Real-time streaming
    Capabilities
      30 languages
      2 billion parameters
      48kHz audio output
    How to use
      Python API
      Command-line tool
      Web demo
    Tech stack
      Hugging Face models
      Pip installation

Things people build with this

USE CASE 1

Build a voice assistant that speaks naturally in 30 different languages without pre-recorded voice samples.

USE CASE 2

Create custom synthetic voices by describing them in plain text, then use them in your app without recording audio.

USE CASE 3

Clone someone's voice from a short audio clip and adjust the speaking style while preserving their unique characteristics.

USE CASE 4

Generate high-quality 48kHz speech in real-time for live applications like video games or interactive chatbots.

Tech stack

PythonDiffusion ModelsAutoregressive GenerationHugging Face

Getting it running

Difficulty · moderate Time to first run · 30min

Requires downloading large pre-trained diffusion model weights from Hugging Face; GPU recommended for reasonable inference speed.

Use freely for any purpose, including commercial use, as long as you include the Apache 2.0 license notice.

In plain English

VoxCPM is a text-to-speech system, software that converts written text into spoken audio. Its main technical distinction is that it skips the usual step of breaking speech into discrete sound tokens, instead generating speech directly as continuous audio representations through an architecture that combines diffusion models with autoregressive generation. The project claims this approach produces more natural and expressive speech than tokenization-based systems. The current version, VoxCPM2, is a 2-billion parameter model trained on over 2 million hours of multilingual audio data across 30 languages. Beyond standard text-to-speech, it supports three additional capabilities: Voice Design (describing a voice in plain text and having the model create it without any reference recording), Controllable Voice Cloning (copying someone's voice from a short audio clip while optionally adjusting the style), and Ultimate Cloning (reproducing every detail of a voice by providing both the reference audio and its transcript). Output is 48kHz audio. Installation is via pip, and the model weights are available on Hugging Face. A Python API, command-line interface, and web demo are all provided. The model can run in real-time streaming mode and is released under the Apache 2.0 license, permitting commercial use.

Copy-paste prompts

Prompt 1
How do I install VoxCPM and generate speech from text using the Python API?
Prompt 2
Show me how to use VoxCPM's Voice Design feature to create a custom voice by describing it in text.
Prompt 3
How can I clone a voice from an audio file using VoxCPM's Controllable Voice Cloning?
Prompt 4
What's the difference between Controllable Voice Cloning and Ultimate Cloning in VoxCPM, and when should I use each?
Prompt 5
How do I run VoxCPM in real-time streaming mode for a live application?
Open on GitHub → Explain another repo

Generated 2026-05-18 · Model: sonnet-4-6 · Verify against the repo before relying on details.