Instruct mode synthesis the instruct text in the final audio

[shuaijiang]

I try the instruct mode to synthesis the audio according to the instruct text, but the audio also contain the intruct text

from cosyvoice.cli.cosyvoice import CosyVoice
import torchaudio
cosyvoice = CosyVoice('iic/CosyVoice-300M-Instruct')
output = cosyvoice.inference_instruct('在面对挑战时，他展现了非凡的勇气与智慧。', '中文男', '女声，快语速')
torchaudio.save('instruct.wav', output['tts_speech'], 22050)

[aluminumbox]

I tried the same code but didn't see the problem, try again

[shuaijiang]

thx, I get the point. The instruct text don't support Chinese?