About the license
Hello, We are planning to use OmniVoice and its fine-tuned models to generate audio for our YouTube channel videos, which is currently monetized through ad revenue. We noticed that OmniVoice utilizes your company's higgs-audio-v2-tokenizer, which includes the following clause:
2. Additional Commercial Terms. If the annual active users of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 100,000 annual active users in the preceding calendar year, you must request an expanded license from Boson AI, which Boson AI may grant to you in its sole discretion, and you are not authorized to exercise any of the rights under this Agreement unless or until Boson AI otherwise expressly grants you such rights.
We would like to clarify a few details regarding this policy:
Scope of the User Limit: Does the "100,000 annual active users" threshold apply to the combined total of all products owned by our entity, or does it apply exclusively to the active users of the specific commercial product built using OmniVoice or the higgs-audio-v2-tokenizer?
Definition of Active Users in Media Production: Since we are solely generating audio for video production and will not be selling any model-related services directly, would the subscribers (or viewers) of our YouTube channel be considered "active users" under this definition?
I have reached out to your company through multiple channels and have never received a response. Given that you chose to open-source your work, why impose such strict licensing restrictions? Furthermore, refusing to explain or respond to questions about the terms is an irresponsible attitude toward the open-source community.
Hello,
Thank you for reaching out, and apologies that your earlier messages did not receive a timely response. We understand the need for clear guidance before using the model in a commercial workflow.
For the 100,000 annual active user threshold, our intended interpretation is that it applies to the product or service that uses the Higgs Materials, or a derivative model built from them, rather than to unrelated products or services owned by the same entity. That said, if the relevant product or service is part of a broader affiliated offering, we may need to consider that context case by case.
For media production use cases such as generating audio for monetized YouTube videos, we generally view the YouTube channel and its audience as the relevant commercial use case, even if you are not selling model access or model-related services directly. Therefore, if the channel or related service exceeds 100,000 annual viewers, subscribers, or other active audience/users in the preceding calendar year, we would ask you to request an expanded license.
Based on the use case you described — generating audio for videos on your monetized YouTube channel using OmniVoice and/or fine-tuned models that depend on higgs-audio-v2-tokenizer — we are open to reviewing an expanded license for this specific use.
Please send the following information to contact@boson.ai so we can document and confirm the license review:
Your legal entity name and country of incorporation, if applicable
- The YouTube channel name and URL
- Approximate annual views and subscriber count
- Whether the generated audio will be used only in your own videos, or also provided to third parties
- Whether you distribute any model weights, fine-tuned models, tokenizer files, APIs, or software based on the Higgs Materials
Once we have those details, we can confirm the expanded license in writing.
We also appreciate your feedback regarding the license terms and our responsiveness. The reason for the commercial threshold is not to prevent normal creative or commercial use, but to ensure that high-scale commercial deployments are reviewed separately, especially where the model or its derivatives may be integrated into products, services, or media workflows with substantial reach. We agree that the process should be clearer, and we will work on improving the licensing contact and response flow.
Best regards,
Boson AI