Commercial use clarification - training data license for VoxCPM2

[Alrescha0312]

Hi VoxCPM Team,

We're planning to integrate VoxCPM2 into a commercial product (a marketing/promotion tool that generates voiceovers from text using cloned user voices).

Before proceeding, we'd like to clarify:

1. What training data was used to train VoxCPM2? Are there any restrictions on commercial use of outputs generated by VoxCPM2?

2. The README states Apache-2.0 license for weights and code. However, some open-source TTS models use training data that restricts commercial use. Does VoxCPM2 have any similar data licensing concerns?

3. For voice cloning features (Controllable Cloning / Ultimate Cloning), are there any additional restrictions when the reference audio belongs to a private individual (not a public figure)?

We're very excited about VoxCPM2's performance and look forward to your clarification. Thanks!

[a710128]

Thank you for reaching out and for your interest in using VoxCPM2 in a commercial product. As we are not legal professionals, the following replies represent our stance and do not constitute formal legal advice:

1. Training Data Under current Chinese legal regulations, there are no restrictions on the commercial use of the data we utilized. However, to protect our trade secrets, we are unable to disclose the specific details of the dataset.
2. Licensing and Commercial Use Our model is free and open-source. It is your responsibility to independently evaluate and ensure that any product or service you develop using this model fully complies with the laws and regulations of your respective country or region.
3. Personal Voice Cloning You must independently obtain lawful authorization for any reference audio used for cloning to ensure that your usage does not infringe upon the legitimate rights and interests of others.

We are delighted that you appreciate the model's performance and look forward to seeing the great products you build with VoxCPM2!