flac vs dsf

[skypickle]

From soundliason.com i obtained the song, A 1000 Shades of Blue by Carmen Gomes Inc in three formats.

Music Scope was used to look at the spectrogram.

Here is the dsf file

 

 

Here is the flac

 

Some differences are quite obvious:

1) the flac goes to 48 khz, the DSF shows a rather abrupt roll off at 22khz and then a broad 'hill' that starts at silence, peaks at -50 db at 44khz and tapers back to silence by 88khz

2) the flac actually has a greater dynamic range with an LRA of 9.1 compared to 8.6 for the dsf

3) the CREST for DSF is greater than that for flac.

 

My questions (due to lack of knowledge and understanding) are:

1) why is the dsf filtered above 22khz?

2) where does that hill come from and why do they leave it in the music?

3) do the frequencies above 22 khz in the flac result in distracting noise due to beats? (for example if there are two tones at 50 khz and 52 khz, there will be a 2 khz beat frequency)

4) The crest factor is the relation between the Peak und RMS values in dB - more CREST= greater dynamic range. The LRA is an indication of music dynamics. (https://www.xivero.com/musicscope-online-manual/)

How can the CREST and the LRA disagree?

 

BTW I also examined the pcm track and it is identical to the flac.

[mansr]

The album was recorded in 96/24 PCM. The DSD download is a conversion. The "hill" you mention is the ultrasonic noise always present in DSD. Apparently Music Scope downsamples DSD to 176.4 kHz with a standard 50 kHz low-pass filter which is why it drops off towards 88.2 kHz.

 

Almost all Sound Liaison recordings are done in 96/24. It's the source for any other formats they offer, so just get that. Their DSD downloads are a waste of money.

[skypickle]

thank you for your reply, but I am still befuddled.

 

why is ultrasonic noise present/added in DSD but not in the pcm?

DSD is 1-bit with a 2.8224 MHz sampling rate. Wikipedia says

'Because of the nature of sigma-delta converters, one cannot make a direct comparison between DSD and PCM. An approximation is possible, though, and would place DSD in some aspects comparable to a PCM format that has a bit depth of 20 bits and a sampling frequency of 96 kHz'

so what does this mean

'Apparently Music Scope downsamples DSD to 176.4 kHz with a standard 50 kHz low-pass filter which is why it drops off towards 88.2 kHz.' I do not understand how you deduced that downsampling is occurring to 176.4 khz from the observation that frequency drops off at 88.2 khz. Why would you not simply say that music scope is down sampling to 88.2 khz?