# OpenBMB/VoxCPM

VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning

Repository: https://github.com/OpenBMB/VoxCPM
Canonical: https://ross.abutalabs.com/products/voxcpm
Homepage: https://voxcpm.com
Language: Python
License: Apache-2.0
License Family: permissive
Topics: audio, deeplearning, minicpm, python, pytorch, speech, speech-synthesis, text-to-speech, tts, tts-model, voice-cloning, multilingual, voice-design, voxcpm
Last push: 2026-08-26T02:42:07+00:00

## Health v2 (maintenance only)
Score: 79/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 83, longevity 25
- inputs: {"age_days": 351, "days_push": 7, "days_rel": 114, "gap_med": 1, "n_releases_24m": 14}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 36145, forks 4132 (observed 2026-08-28T04:12:00.973061+00:00)

## What it is
VoxCPM is a tokenizer-free text-to-speech system built on a diffusion autoregressive architecture that generates continuous speech representations end-to-end. VoxCPM2 is a 2B-parameter model trained on 2M+ hours of speech, supporting 30 languages, creative voice design, controllable voice cloning, and 48kHz output.

## Use cases
- generate natural speech from text in multiple languages
- clone a voice from a short audio sample
- design a custom voice with controllable characteristics
- produce studio-quality 48kHz tts audio
- build a voiceover or audiobook generation pipeline
- run zero-shot speech synthesis locally in python

## When to choose
- you need expressive, natural-sounding tts without discrete tokenization artifacts
- you want multilingual synthesis across 30 languages
- voice cloning or creative voice design is a core requirement
- you want a pip-installable python tts model with a permissive Apache-2.0 license

## When to avoid
- you need a lightweight tts for CPU-only or embedded devices
- you need real-time streaming synthesis with very low latency
- you lack a GPU and cannot afford heavy inference compute

## Facets
- artifact type: library
- maturity: active
- function: tts, speech-recognition, machine-learning, deep-learning, audio-processing
- domain: speech-processing, machine-learning, deep-learning
- platform: python, cross-platform, cli
- tags: text-to-speech, voice-cloning, voice-design, tokenizer-free, diffusion, multilingual, pytorch, minicpm, zero-shot-tts, audio, natural-language-processing, gpu

## Member repositories
- OpenBMB/VoxCPM (main) score 79

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:00.973061+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:26:43.380277+00:00, confidence not recorded.
  - readme: https://github.com/OpenBMB/VoxCPM (fetched 2026-08-28T04:12:00.973061+00:00, sha 0bbbe95f445d)
  - homepage: https://voxcpm.com (fetched 2026-08-29T07:47:42.551575+00:00, sha 3e6b11236e63)
  - registry_pypi: https://pypi.org/pypi/voxcpm/json (fetched 2026-08-29T07:47:42.562266+00:00, sha 2cf8a9625e43)
- Data as of 2026-08-30T08:39:29.467469+00:00.