stepfun-ai

/

Step-Audio-2-mini-Base

like 22

Follow

StepFun 1.49k

Hi, Step-Audio-2-mini-Base is the base model for Step-Audio-2-mini and it aims for end-to-end speech conversation.

However, the base model should also have some zero-shot voice cloning ability, by prefilling the prompt text-audio interleaving tokens and completing new tokens based on given text.

This usage is not included in our examples.py.

hey @petronny , is it possible to implement voice cloning in an example? Or a Google Notebook Colab example to check the voice cloning capability? that would be so helpful and I also want to ask is it possible to clone a voice by finetuning or training the model?