The output says resampled from 44 to 384. I have 3 different systems with 3 different Dacs. My main system uses a Teac Dac with Magnepan speakers. My second system uses a Ps audio Dac with B&W speakers. On both of these systems using MC21 I notice a sound degrading, not as clear or detailed sound. The sound degrading is worse with the Teac Dac. Both systems sound excelent using MC 19. I will reinstall MC 21 if this will improve the audio playback. At this point, I feel that there was change in MC 21 that affects the audio quality. Has anyone else noticed this changes. I will continue to use MC 19.
So, everything you say so far points in the direction of subtle differences, reducing the probability of something broken.
But you still haven't really
documented that the audio path is the same for the two versions. And I would point out again that you should switch off as many intermediate steps as possible (such as upsampling) when doing the first comparison, then add one by one and test in each step.
Also, if Volume Levelling is on, and MC19 uses Replay Gain and MC21 is using R128, audio level will be different and higher level is almost always percieved as 'better sounding' (i.e. you will not be comparing apples to apples).
I think devs or someone with detailed tech ability and good memory will need to tell what has happened to the audio engine between 19 and 21. What comes to mind, but I do not remember the time of implementation for all of these, are:
- 32 to 64bit audio engine (when?)
- Changes on how playback from memory is handled (when?)
- Different dithering (TPDF has just been added to MC21)
- other things?