Hey everyone, I’m using VEO3.1 to clone voices. I prefer VEO3.1 over ElevenLabs because the results feel higher quality - more expressive, and better synced with the visuals. The drawback is that the voice doesn’t match the avatar owner’s voice exactly. To keep the voice consistent across generated videos, I use a detailed voice description - I call it a Voice Profile. So I built an agent that generates Voice Profiles from a photo or a text description of a person, and then I use that Voice Profile inside VEO3.1 as a part of the prompt, along with script for avatar. I’m curious if anyone else has tried a similar approach - would love any feedback or tips. https://chatgpt.com/g/g-690e0cec08b88191a5ac5a1d37e8446a-character-voice-composer