VoxCPM
Tokenizer-free TTS built for expressive, context-aware speech.
About
VoxCPM is a text-to-speech system from OpenBMB that skips discrete speech tokens entirely and generates continuous speech representations with a diffusion-autoregressive design. It is built on a MiniCPM-4 backbone and trained on a reported 1.8 million hours of bilingual audio, aiming for expressive prosody that adapts to the content plus zero-shot voice cloning from a short reference clip. Weights for VoxCPM-0.5B were open-sourced in September 2025, with a 1.5 version following in December. The license terms were not confirmed in the sources read, so check the model card before any commercial use, and independent quality comparisons are still thin.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Weights open-sourced 2025-09-16 (0.5B), technical report 2025-09-29, 1.5 weights 2025-12-05; OpenBMB is an active maintainer (MiniCPM family); License terms not confirmed in sources read; verify on the HF model card before commercial use; 1.8M-hour training corpus and quality claims are author-reported; independent comparisons thin. License: License not confirmed from sources read; check the openbmb/VoxCPM-0.5B card before use..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 6.0/10
- Manual review · sanity read + red-flag check
- 6.0/10
- · Weights open-sourced 2025-09-16 (0.5B), technical report 2025-09-29, 1.5 weights 2025-12-05
- · OpenBMB is an active maintainer (MiniCPM family)
- · License terms not confirmed in sources read; verify on the HF model card before commercial use
- · 1.8M-hour training corpus and quality claims are author-reported; independent comparisons thin