This feature lets you quickly create a voice clone model using your own audio recordings. You can then use this cloned voice to generate audio using the Cloud TTS API, supporting both streaming and long-form text inputs. For safety reasons, Brand Voices Lite is only available to allow-listed users. To access this feature, contact a member of the sales team to be added to the allow list.
Obtain consent
This pre-GA feature shouldn't be utilized in production. Your team is required to complete a safety and ethics review to obtain approval before using this voice in production. Under no circumstances should this feature be used for cloning voices without explicit consent from the voice talent.
Responsible AI requirements
We are dedicated to upholding the highest standards of integrity and ethics. The Brand Voices feature requires explicit consent from the speaker for every voice created. A recorded statement from the original speaker must confirm their consent. When you submit the audio files, we perform a consent check to ensure proper permissions are in place. We also review the intended use cases before granting access.
If your training data doesn't pass our internal verification and validation check, you might need to rerecord or resubmit the data after fixing the identified issues. For further assistance, contact the sales team.
What's next
Implement the Brand Voice Lite feature using our quickstart.