Hi, I’m new with Audrivana. I’m wondering why the Leveling is done after Upsampling but not before it wen I choose Upsampling to DSD. It’s well known, then it is not possible to change the volume in DSD format.
So probably it is needed to re-convert to PCM, change the volume and then convert to DSD again. Wouldn’t it be better to do the Leveling first on the PCM signal and the convert it to DSD?
Might be there is somewhere the answer to my question but I wasn’t able to find it.
Welcome…
The PCM file gain can be reduced globally prior to upsampling in the settings window… You can find more information about particular settings in the Online User Guide pop-up windows found under the (i) icon at the top-right of the settings window…
From the Upsampling Settings User Guide:
To avoid problems caused by oversampling of DSD, which can cause interruptions during audio playback, it is advisable to reduce the sound volume. For tracks that use ReplayGain and for tracks with sampled peak values, the volume reduction is calculated automatically. For other tracks, a precautionary volume reduction can be specified here at the bottom of this settings window.
Hi, in my understanding it has noting to do with “The PCM file gain can be reduced globally” because I want to use the Leveling-step playing Playlists e.g. from TIDAL to get nearly the same loudness for each song dynamically. so global settings won’t work for thet. It makes more sense to do the Leveling yet in PCM format before conversion to DSD. So why Audrivana do it in the reverse order?
Typically DSD gain is adjusted in the analog domain at the preamp or in my case in the headphone amp… Folks like me don’t want to do any leveling on any files… However, you can employ dynamics processing in the plug-ins module, which is processed prior to upsampling.
Sorry, I’m afraid, we talk past each other…
Playing a Playlist one can’t know the loudness of the next songs. So it’s needed to switch the Loudness Levelling on to equate the loudness. The Levelling has to be done in PCM format. So in my opinion it would make much more sense to do the Levelling first and then the conversion to DSD.
My simple question is: why Audrivana do it in the reverse (in my opinion wrong) order?
Maybe somebody who understand exactly my question could explain me that…
Reducing the gain or leveling the gain of a PCM file, requires decimation/discarding of bits (dynamic-range) calculated as: (±) 6dB/bit… Some folks don’t care about losing bits in the process, some do… I am one that wants no alteration to the source file(s) dynamic-range prior to modulation to DSD… It’s my opinion as one that modulates all PCM files to DSD128, that it is a mistake to do otherwise, if one appreciates the results of the modulation in the 1-bit DSD playback.
Upsampling lower resolution PCM files to higher resolution PCM and subsequently applying Volume or Leveling DSP, retains the source file dynamic-range by calculating at 64 bit and the result is output as 24bit or 32bit files. So it is increasing the dynamic-range, not decreasing the source file dynamic-range.
Hi, the explanation is that Audirvana does not “do it in the reverse order.” In spite of language that might cause confusion, volume reduction is applied before conversion to DSD, and this conversion is done only once, not back-and-forth.
I hope this answers your question.
Are you extrapolating from the statements made in the online User Guide?
Because Volume adjustment of the resulting 1-bit PDM signal is converted to DXD 24/352.8kHz to apply volume DSP and modulated again to the 1-bit PDM sample-rate before output, if applied.. Are you specifically speaking about the Leveling of files?
The Replay Gain function allows you to automatically adjust the volume of tracks to maintain a consistent playback level within a playlist. This adjustment is made according to a sound level reference specific to each song according to EBU R128 and DR standards.
If this reference is not available in the metadata of a piece in your local library, Audirvāna can calculate it.
You can start the analysis of the albums in your library by pressing the ‘start’ button. You will then open a view that displays all the albums that do not have this information. You can select all or some of them, and launch the analysis in background.The Replay Gain can also be applied to songs streamed from a partner service, provided that this service makes this information available (e.g. TIDAL).
It is advisable to start with the “Preserve album dynamics” option which preserves the dynamic variations desired by the artist within an album.
Hi Jud,
thanks for the answer. If it is so, it’s good. That’s what I would expect and what makes sense in my opinion.
But the screenshot of Audrivana
shows the opposite. The step Volume Levelling follows the step Upsampling, which is confusing . So what’s the truth?
Upsampling is done in PCM. The conversion to DSD is done by something called “delta-sigma modulation” or “sigma-delta modulation” (two names for the same thing). The DSD specification calls for -6dB reduction after upsampling and before conversion to DSD, but in practice -3dB or so is probably adequate. Audirvana gives you a choice of what level of reduction you would like to apply.
Since this volume reduction is applied in the middle of the upsampling and modulation process that we informally call “upsampling to DSD,” how should it be visually represented without being overly complicated? What you see is the visual representation Audirvana has chosen.
To avoid being confusing, the two steps, in my opinion, should be shown in opposite order:
first: Volume Levelling with PCM input and still PCM output
following by : Upsampling, in this case conversion PCM to DSD by delta-sigma modulation with DSD output
If it was already so, what in my opinion makes more sense an probably is closer to the truth, I wouldn’t ask this question.
Volume Levelling with input DSD (output of the previous step) and output DSD, like on the screenshot above, is misleading and leads to the conclusion that there is another DSD to PCM and PCM to DSD conversion inside this step needed. I think/hope such change in the order of the displaying these 2 steps would be very helpful and technically understandable.
In the SDM where the LPCM file is interpolated (modulated) to 1-bit PDM (DSD) the sample-rate is increased and the dynamic-range is determined by the nature of the noise-shaping… Why would there be reason for an intermediate and superfluous interpolation to a higher resolution PCM bit-depth and sample-rate before modulation to 1-bit PDM…? It doesn’t happen…. The SDM architecture determines the interpolation algorithm.
I’m not convinced that any dynamic-range/level/amplitude reduction DSP is applied to the PCM file(s) prior to interpolation/modulation to DSD outside of the SDM settings… You would need to show clear evidence of this being so, or we get clarity from the Audirvāna Team… It makes sense to me that Volume and Leveling DSP would be be applied post interpolation of the PCM file(s) to a higher PCM sample-rate or DSD. because the dynamic range of the source file encoding will not be affected by the truncation of bits in those DSP actions applied to the SDM output file…
For example PCM 16/44.1kHz modulated → DSD128 will be decimated to DXD 24/352.8kHz for volume DSP and subsequent of the volume/Leveling DSP, modulated to DSD128 for output… There is no loss of resolution or dynamic range of the source PCM file(s) encoding… just reduction of the dynamic-range/level of the DXD 24bit, 352.8kHz file that encapsulates the source PCM 16/44.1kHz encoding which remains intact.
Maybe I’m answering my own question here…
The question is centered about where the interpolation to 24/352.8kHz happens in the DSP path and if the signal routing is dynamically switched based on Volume/Leveling settings…
In other words, the multi-bit DXD file becomes the reference output level that gets modulated to 1-bit DSD… The DXD file has plenty of headroom/dynamic-range to dispose of…
Either way… there is little to be concerned about regarding where in the DSP chain the Volume/leveling processing occurs…