BreezeBlue models, access and benchmark guide
BreezeBlue model and access guide with a practical US-English evaluation brief and links to publisher materials.

👁 111 reads
For a customer installation, an available Breeze-TTS checkpoint is the beginning of the procurement question. Read the repository license, determine the target hardware and separate its terms from those of a commercial hosted service.
BreezeBlue released the Breeze TTS 2 weights and PyTorch inference code. Its license is explicitly for research and non-commercial use. A university laboratory evaluating how the voice handles a place name has a different permissions question from a business installing the same checkpoint on a customer helpline. For a commercial deployment I would obtain the applicable permission before quoting hosting costs. The two rankings below are also different purchases in practice. One lets the provider choose a voice. The controlled-voice track constrains that choice; its lower score tells the buyer to audition the actual voice the installation requires.
Choosing the first job
Synthesize a consented sample with a hyphenated place name and record which revision and license produced the result.
Models in the November 17 snapshot
| Model and track | Placement | Reported score |
|---|---|---|
| Breeze TTS 2 (AA provider voice) | 14 | 1229 Elo |
| Breeze TTS 2 (AA controlled voice) | 31 | 1019 Elo |
Release record. Publisher documentation identifies the product family. The ranked configurations include Breeze TTS 2.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →Sources
- https://huggingface.co/BreezeBlue/Breeze-TTS-2 (accessed November 17, 2026)
- https://huggingface.co/BreezeBlue/Breeze-TTS-2/blob/main/LICENSE (accessed November 17, 2026)
- https://artificialanalysis.ai/text-to-speech/leaderboard/provider-voice (accessed November 17, 2026)
- https://artificialanalysis.ai/text-to-speech/leaderboard/controlled-voice (accessed November 17, 2026)
More on this from QUANTSLANT
benchmarks·AssemblyAI models, access and benchmark guideAssemblyAI model and access guide with a practical US-English evaluation brief and links to publisher materials.