Hello, I am evaluating AudioWeaver on the STM32F407 Discovery platform (eventually I want to run it on the STM32L4 series).
My goal is to evaluate/estimate the power consumption of the DSP processing.
I implemented a system as in the attached image and observe 30% CPU utilization and everything sounds great, however when I add another graphic eq on the other channel I see 60% utilization, but the audio does not sound well. Anything that I could be doing wrong? Please advise.
I am assuming that if the CPU % at Audio Weaver Server shows about 30% that means that the duty cycle is 30% and the power consumption will be whatever the CPU power consumption for 30% in active mode, and 70% in idle mode. Is this correct?
Thank you.
1:00pm
Hi pavo28,
The CPU load calculated in Audio Weaver is only the time it takes for the audio processing. That means that things like processing of IO and IRQs are not included in this measurement.
In the ST BPSs, a demo version of our proprietary ASRC is included on the USB and microphone input streams. The ASRC manages the problems that arise from clock drift between the audio source and the audio processor, but it is computationally expensive. What you're hearing is audio blocks being missed because audio processing + ASRC + other IO > 100% of the available CPU.
For a more detailed answer, and some ways to mitigate, see this post: https://dspconcepts.com/forums/audio-weaver-designer/140-audio-weaver-61...
Thanks
-Axel