I tried applying HD Audio analysis on some tracks that I have down-converted from DSD64 to 24/352.8kHz and it shows them as having a Nyquist frequency filter cut-off of approximately 22.7kHz… How do I know if this analysis is correct? The files are being read as 24/352.8kHz and the Nyquist frequency filter cut-off should read approximately 176.4kHz … I understand there may be a possibility these files were mastered as native 16/44.14kHz then subsequently unconverted to DSD64. I really don’t know if these files were mastered as DSD64… Is the analysis being thrown-off by my down-conversion to 24/352.8kHz or is it really identifying the native master resolution that created the DSD64 product? Does the HD Audio analysis try to only show the Nyquist of the original master and not the current sample-rate? It would be nice to be able to toggle between the track playback Nyquist analysis and the analysis of what the original source master might be…
Can you show the analyser result with a printscreen?
Do you mean that you converted the files to PCM with an external converter? Or that you let Audirvana do this Realtime. I suspect the analyzer is showing the information from the file and not the result of processing from Audirvana.
I wouldn’t describe it as a down conversion from DSD 64 to PCM352.8. DSD64 will have to be filtered above 25khz. So the 22.7kHz isn’t that strange. In the same range as Redbook (16/44.1)
What do you expect above 20khz? Most instruments don’t get there. For many mics this is above range, mic preamps don’t come up to 176.4khz. Neither do AD converters. What kind of energy or information should be above 30kHz?
I down-converted (decimated, remodulated) DSD64 (2.8MHz) files to PCM 24/352.8kHz files and HD Audio analysis read the 24/352.8kHz file playback… The theoretical Nyquist frequency of a 352.8kHz PCM signal is approximately 176.4kHz, where it hits a brick-wall filter for playback… HD Audio analysis does not report graphically the sample-rate and Nyquist cut-off of the file currently playing. The process only makes an educated ‘guess’ as to what the original “mastering sample-rate” might have been…
In regard to A/D and sample-rate…How do you think native DSD64, DSD128, DSD512 and DSD1024 (2.8MHz, 5.6MHz, 11.2MHz and 22.4MHz) recordings are converted to digital? This is not the place to get into the neuropathology of human auditory perception, head transfer function and psychoacoustics, nor the technical limitations of recording systems.
The HD analysis tells in which frequency spectrum energy is present. I don’t think anything more or less. So I don’t quite understand what exactly you expect from the analyzer.
In this case I’m questioning the validity of the analysis and whether or not the decimated DSD64 files were really DSD masters or did my offline remodulation/decimation, confuse the algorithm making it difficult to determine the ‘mastering’ sample-rate… We cannot really compare DSD with PCM, so I’m wondering what was the HD Audio analysis actually analyzing…
Ok… I get what you are saying in regard to the actual energy spectrum… This would make sense, because these recordings were originated from analog tape… Thanks for the reframe…
Why won’t you apply the analysis to the original DSD64 files to see what AS says about them?
Is this possible ?
Doesn’t work on DSD file playback
DSD is not a great format for editing or analysing. It works wel for archiving.
I don’t think you should expect too much from this tool. It is not possible to trace back what a master was based on the converted file.
A hi res file should have energy beyond -96 dB and 44.1kHz. If you see a clear brickwall there then it is likely that the master was a 16/44.1. A DSD64 does not have much greater range than a 16/44.1.
There are also plenty of tools to download that allow you to do the same and more. I use MusicScope
Yes… I understand the formula for LPCM dynamic range: 6 x (number of bits) + 1.75… You cannot compare 1-bit Sigma-Delta modulation to Linear Pulse Code Modulation… the filter requirements are completely different as the noise gets pushed-out with sample frequency…
MusicScope should be able to analyse DSD.
I’ve never tried it.
Edit:
Apparently MusicScope is no longer for sale.
I didn’t look any further for alternatives.