OpenBMBVoxCPM
Tokenizer-free multilingual TTS with creative voice design and true-to-life cloning.
TL;DR · 30-second scan
VoxCPM (Python) — Tokenizer-free multilingual TTS with creative voice design and true-to-life cloning.
You are starting a new TTS project in 2026 and want the best open baseline available.
Voice & Audio Generation
The strongest open TTS released in 2025-2026. Tokenizer-free architecture means better prosody than SoVITS on long text. Voice design lets you generate a fresh voice from a description instead of cloning one. Apache-2.0 license and active maintenance from OpenBMB. Setup is easier than SoVITS but the models are larger. Currently the best default pick for new open-TTS projects.
You are starting a new TTS project in 2026 and want the best open baseline available.
You have already tuned GPT-SoVITS to a specific voice and it works — the switching cost is not worth it.
Add this badge to your README to show your project is curated on StackPicks. Free, lightweight (180×28 SVG), and gives your visitors a one-click way to see honest take + alternatives.
[](https://stackpicks.dev/repo/voxcpm)
<a href="https://stackpicks.dev/repo/voxcpm"><img src="https://stackpicks.dev/api/badge/voxcpm" alt="Featured on StackPicks" width="180" height="28" /></a>
Are you the maintainer of OpenBMB/VoxCPM? Add the badge and we'll feature your project in the weekly curator newsletter.