Bias controlled testing of CPU load on audio

[plissken]

I want to offer a passively cooled, headless, remotely controlled, Windows 10 PC that can be used to drive members audio stack from their USB cable on back.

 

This is to gain data on listeners ability to detect when a CPU has load on it vs not.

 

Looking for input on how best to accomplish this but here are my thoughts:

 

1. Use JRiver with it's remote function based on a tablet/phone that I could include.

2. Music samples from 2NL 24/192 PCM on local storage

3. Use Geekbench 5 to run a mixed application load

4. Random generation of Minute and Second for the 80% middle spot of the song for a change. I.E. if a track is 10 minutes long changes would have to happen after 1 minute and before 9 minutes

5. Participant can listen to the system in it's default state of minimal load to acclimate to the sound signature.

6. Get enough participants that the setup can be forwarded on directly

7. No access granted directly to the computer by the listener.

8. Document the screen via recording software so everyone can see when the changes are made and watch the process monitor

 

I would include a temporary SSID and passphrase that a member could stand up and replicate with their WiFI to onboard this for remote headless operation for the duration.