KPU and audio record incompatibility

I’ve created a cnn model for sound classification which uses fft images, but when I have the “audio = rx.record(sample_points)” on I can´t use the forward() function from the KPU because the board just crashes without any error message (sometimes it shows half a message that starts with 0;33mW or something)
I’m on a time schedule to deliver the project so any help is appreciated.

edit: when I run the script on the terminal after transfering it to the board returns this error message:
“W (129088164430) SYSCALL: sys_exit called by core 0 with 0x1”


seems like a similar issue I have: