Dynamic Comparison of CD, DVD-A, SACD - Part 1 - page 2
These are plots of the "density" of frequency content across the 0-48kHz spectrum. The brighter the colour, the greater the amount of frequency content at any point in the graph.
First of all, the EAC rip:
Note the total absence of any frequencies above 22kHz, as to be expected given the 44.1kHz sampling rate.
Now, this is the same CD track but as played by the Sony and captured using the Prodigy:
Interestingly, various points in the track are "corrupted" by high frequency "spikes". I am not sure what is generating this noise, but it could be a combination of the player, the amplifier and the sound card.
Here is a frequency analysis at 4:29 into the track (at the point represented by the yellow dotted line in the graph above):
Notice, again, the severe drop off above 22kHz. In comparison, the EAC rip at around the same point is similar, but without the ultrasonic noise:
It looks like Sony has opted for a gentler filter roll off (probably achieved using upsampling).
By comparison, this is the spectral analysis for the DVD-A:
Note that we have true frequency content above 22kHz all the way up to around 36kHz. The spikes above 36kHz are probably the same "noise" found on the CD version, although I cannot confirm that. The purple horizontal lines around 36kHz is probably the analog tape bias.
The frequency analysis around 4:29 for PCM looks like this:
Again, this confirms the presence of real musical content above 20kHz, but dropping off fairly significantly past 30kHz. I suspect there is no real usable musical information above 30kHz.
In comparison, this is how DSD looked like:
As you can see, DSD is almost as good as PCM in capturing ultrasonic frequencies, but at the cost of higher ultrasonic noise (note the purple specks above 35kHz).
Frequency analysis at 4:29 :
As you can see, DSD offers a response very similar to PCM except for the overlay of the ultrasonic noise above 30kHz.
Comparing the wave files between 2:50-4:50 yielded a surprise. I put the CD, DVD-A and SACD versions side by side and you could really see the difference
The CD is the most "compressed" (average signal amplitude higher in comparison to the peak) and DSD the "least" compressed. This means that if you adjust the relative levels of the three recordings such that the relative "energy" across all three are approximately the same, then the peaks and transients in SACD will be "higher" over DVD-A, which in turn will be higher than CD.
The differences are hard to see on this web page because I have shrunk the waveforms, but on Cool Edit they were quite obvious, and represents around a 2-3dB difference, which is significant enough to be audible. This difference was also noted empirically by me during recording, as I had to lower the gain for the CD and increase the gain for SACD.
If this difference is "real", as opposed to an anomaly in my equipment, then it could explain why some people don't like SACD compared to CD or DVD-A. The slightly lowered sound levels, if not compensated during the listening process, will cause SACD not to sound "as good" compared to CD or DVD-A.
The better dynamic performance of SACD would also explain why some people prefer SACD, as they probably notice the slightly higher dynamics.
The results are interesting indeed, even though I would caution against over-generalizing them into conclusions about each format. Remember that the results may not be applicable beyond a single title and the constraints of my equipment.
© 2004 Christine Tham
Reprinted with the Permission of Christine Tham. Please visit her very informative website at: http://users.bigpond.net.au/christie/