Instructions to use BreezeBlue/Breeze-TTS-2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use BreezeBlue/Breeze-TTS-2 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-to-speech", model="BreezeBlue/Breeze-TTS-2")# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForSeq2SeqLM model = AutoModelForSeq2SeqLM.from_pretrained("BreezeBlue/Breeze-TTS-2", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Breeze-TTS-2 license concern: GGUF distribution requires user acceptance
Hi BreezeBlue team,
I didn’t really pay attention to the license terms until @a2industries brought this up. One clause concerns me:
“Ensure that each recipient independently accepts this Agreement before accessing or using the Model Materials or Derivative Model.”
The license explicitly treats quantization as a Derivative Model, so this requirement would also apply to many GGUF repos. The key issue is that the obligation is placed on the distributor to ENSURE each recipient accepts the agreement before getting access, which is difficult to satisfy with normal public HF repos or automatic model downloads. This would make normal open-source distribution and integrations difficult.
Licenses of other popular models such as Fish Audio, Higgs Audio TTS, and MiniMax-Music 3 don’t impose this kind of distributor-side acceptance requirement; they generally rely on the recipient being bound by the license through access or use.
@audio-cpp
Thanks for pointing this out. This is a valid concern.
Our intent was to ensure that downstream distributions remain subject to the same research and non-commercial terms, not to require every distributor to implement a separate click-through acceptance mechanism.
We have updated the License to reflect our original intent.