Likely to be dependent on the complexity of your code, and what it's doing
Is it something that you can port to the single-precision FPU of the CM4F (Cortex-M4) used in several STM32 models? If you can establish a port to the ARM core, you'll be in a position to choose several possible MCU, so would be worth reviewing.
https://www.analog.com/en/products/adsp-2171.html#product-overview
