Comparing mpd to pcp

[Zombie]

"We can hear everything we measure, but we can't measure everything we hear. Let your ears be your guide."

Also known as the snake oil theorem

[slartibartfast]

Hmm I might not have set the volumes to 100% [emoji2357]. I'll try again...

Sent from my Pixel 3a using Tapatalk

[slartibartfast]

I think you indeed have to do this manually. After that maybe the bit-compare extension of Foobar2000 can be useful, because it is able to correct for small differences in alignment afaik.

I connected the output of my Topping E30 DAC to a Behringer UFO-202 USB audio interface and recorded the same track played with mpd and pcp. I made two recordings of each then compared them with Deltawave after trimming all the recordings to the same length and aligning them as close as possible. Deltawave is meant to correct any alignment errors and gain differences.
The correlated null depth for the comparison between the two pcp recordings pcp1 and pcp2 was 67.22dB
The correlated null depth for the comparison between the two mpd recordings mpd1 and mpd2 was 72.77dB
The correlated null depth for the comparison between pcp1 and mpd1 was 62.31dB
The correlated null depth ideally should be as high as possible.
One unexpected observation was that piCorePlayer was quieter than mpd by 3.14dB with replaygain disabled. How could that be explained?
Just in case anyone is interested the full results are available at this Dropbox link.

[slartibartfast]

I think you indeed have to do this manually. After that maybe the bit-compare extension of Foobar2000 can be useful, because it is able to correct for small differences in alignment afaik.

I did get better results after upsampling the waveforms and manually aligning in audacity but I can still see that the alignment isn't perfect. Programs like Deltawave and Audio Diffmaker claim to automatically align even if samples are not perfectly aligned. I'll try making two recordings of the same player and see how that looks. If I can't get decent results from that I'll give up.

Sent from my Pixel 3a using Tapatalk

[zordaz]

I think you indeed have to do this manually. After that maybe the bit-compare extension of Foobar2000 can be useful, because it is able to correct for small differences in alignment afaik.

[slartibartfast]

You often see claims that one bitperfect player sounds different to another bitperfect player. I thought I would try to analyse the outputs of mpd and pcp by capturing the waveforms using Audacity. It seems to me that the biggest obstacle is aligning the waveforms. Obviously higher sample rates can be more closely aligned but I am limited to 48kHz.
Is it possible to achieve perfect alignment? I don't think it is but thought I'd ask anyway [emoji3]

Sent from my Pixel 3a using Tapatalk