This is the server for the UltraEval-Audio project.
git clone https://github.com/XavierJiezou/Step-Audio.git
cd Step-Audio
conda create -n step-audio python=3.11.7 -y
conda activate step-audio
pip install -r requirements.txtpython download-modelscope.pyCUDA_VISIBLE_DEVICES=4 HF_HOME=./checkpoints MODELSCOPE_CACHE=./checkpoints python adv_api.py \
--tokenizer_path ./checkpoints/stepfun-ai/Step-Audio-Tokenizer \
--tts_path ./checkpoints/stepfun-ai/Step-Audio-TTS-3B \
--llm_path ./checkpoints/stepfun-ai/Step-Audio-ChatNow, you can use --model step-voice in the UltraEval-Audio.