Converting DSD Files to FLAC

[LuvMyQuad]

I have read a lot of par4ken's posts over the years. He's a pretty straight shooter when it comes to describing what he hears. When he first stated his DSD vs PCM remarks in the original "Holy Cow" thread, I too thought there might be some bias afoot, especially since there was no true A/B testing involved, and in view of my experience with comparing DSD and DSD converted to PCM not agreeing with his observations.

I'm coming to the conclusion that he may well be describing the reality of the situation correctly. He is speaking about an audible difference when using an on the fly conversion of DSD to PCM within his Oppo player. Something that would be harder to evaluate A/B style.

Early on in my SACD ripping education, I tried to test for audible differences in DSD vs DSD converted to PCM. I wanted some reasonable A/B method. I ended up comparing the following:

Optical SACD playback through my Oppo BDP-83 SE using its internal DAC and inputting analog audio to my pre pro
VS.
A 24/88.2kHz PCM conversion of the same tracks played via Kodi on my HTPC with the HDMI sent to the same pre/pro and using its internal DACs for decoding.

I tried to use predominately 4.0 SACDs to eliminate any issue with LFE level settings.

This way I could set the two sources to start playing at more or less the same time. I like to have a 3-5 second delay between them so I can hear a passage and then quickly re-play the same passage from the other source. I could set levels for each source separately and switch between the sources at will. This is the same setup I have used in the past to compare vinyl to digital, CD to Hi-Res, etc.

Result: I could not reliably perceive a difference in sound. Nor could the much younger ears of my son when he tried it. I know it wasn't a blind test. I know the levels were set subjectively. But it was enough for me to go ahead and feel good about converting all my DSD rips to PCM. (I kept all the DSD rips just in case).

What par4ken is talking about is different. He is referring to his Oppo using its internal DAC to convert SACD to PCM. This is not something I tested. It may well be that this process is flawed in some way. Hence, I cant refute it, and I'm willing to believe it until it is proven wrong.

I do want to get to the bottom of it though. And I also want to know I'm doing all i can to get the best PCM conversion from DSD sources possible.

 

[par4ken]

Bias can be triggered in many ways you'd never expect. As long as you know what's playing it can rear it's ugly head.

That is ridiculous! On several occasions now my Oppo was playing SACD as PCM. I very quickly heard the difference, stopped the playback a switched back to DSD. The difference is very audible!!!! I didn't know it was set to PCM for sure until I looked at the display and sure enough. Sorry but I trust my ears!

 

[Kal Rubinson]

That is ridiculous! On several occasions now my Oppo was playing SACD as PCM. I very quickly heard the difference, stopped the playback a switched back to DSD. The difference is very audible!!!! I didn't know it was set to PCM for sure until I looked at the display and sure enough. Sorry but I trust my ears!

However, that "test" does not tell you on how many other occasions, you were unaware that it was playing as PCM.

 

[LuvMyQuad]

That is ridiculous! On several occasions now my Oppo was playing SACD as PCM. I very quickly heard the difference, stopped the playback a switched back to DSD. The difference is very audible!!!! I didn't know it was set to PCM for sure until I looked at the display and sure enough. Sorry but I trust my ears!

When you have time, it would be interesting if you could compare the converted PCM vs the straight DSD levels out of your Oppo when it plays each one. See if one sounds louder than the other at the same volume level on your prepro/AVR. If you have an SPL meter you can use, even better.

 

[par4ken]

However, that "test" does not tell you on how many other occasions, you were unaware that it was playing as PCM.

None!

 

[par4ken]

When you have time, it would be interesting if you could compare the converted PCM vs the straight DSD levels out of your Oppo when it plays each one. See if one sounds louder than the other at the same volume level on your prepro/AVR. If you have an SPL meter you can use, even better.

Yes I could try to do that. Because level is dynamic, changing with the music signal it might be hard to check for level differences. A test disc with some steady state tones would be useful. Level differences do not account for the difference in sound that I'm hearing however. I'm not flipping back and forth (A-B testing) to hear differences between the two, volume is set to a comfortable level either way. The overall sound quality is the same whether played soft or loud. The only other thing to try would be to make sure that all processing done with the PCM signal is disabled, including mixdown to see if that changes anything.

Subjectively speaking native DSD has a harsher sound, but seemingly more dynamic. The converted PCM sounds tapeish, lacking the punch and fine detail of the native DSD. At the same time PCM from Blu-ray Audio sounds fine, very detailed but with a seemingly smother sound character than SACD. My comparison to Blu-ray is obviously with a different music selection so may not be 100% valid, but represents my generalised impression of the sound quality of both.

 

[marcb]

It seems to me that what may be forgotten in this discussion is, of course, not all DACs are equal - but also individual DACs don’t necessarily do an equal job of different conversions.

Depending upon what DACs one is using, one could get appreciably different results.

 

[sjcorne]

If so seems there may be a slight difference in mastering for the Front channels

I thought it was the HFPA Blu-Ray that was said to have a unique mastering? I only have the SACD.

 

[J. PUPSTER]

I thought it was the HFPA Blu-Ray that was said to have a unique mastering? I only have the SACD.

That's what I was hoping to find also, and that someone would chime in about the levels on the DVD-A GBYBR song; so I don't have to re-rip it.

But from my notes those fronts are ~1db less than the others on the SACD and the same on the DVD-A I currently have. Hope that's not confusing or irrelevant (could be )

 

[ssully]

Your AVR converts to PCM so that your DSP functions can work, unless you output analog from the Oppo (converts DSD to analog) into your amplifiers analog inputs. My systems are all analog, audio processing in my system doesn't require conversion to PCM.

I can see a real schism coming to audio (if it's not already here) between those who love their DSP effects, room correction ect, and the purists. The purists would tend to be more in the stereo only camp, valuing sound quality above all else, using analog and SACD in native form and a minimum of analog signal processing. I find myself more in the purist camp but extended into the surround world, primarily quad/four channel only.

If purists were commonly obsessed with room acoustics and speaker accuracy, the two biggest impacts on home system sound quality, I would believe this to be a schism with a rationale (albeit weak*) behind it. But if the sound from a purist rig at a listening seat is highly colored by the speakers/room -- as is typically the case -- and that coloration can be more or less corrected by DSP, which one is offering higher 'sound quality' in the end?

Also, AVRs that have a DSD decoding chip in them always offer one or more 'purist' modes of playback that bypass DSP. To exploit this option, purists can use an HDMI connection to let the AVR receive the raw DSD bitstream from the disc/file player. Then there only is a single D/A conversion directly from DSD. You may lose bass management.

Alternately the DSD can be directly converted to analog in the player, and then passed via multiple RCA connectors to a preamp/amp that apply no DSP. AKA cutting edge circa the year 2001. To remain virginally analog and still have bass management , you'll need something like the old Outlaw outboard multichannel crossover box.


*worries about converting DSD to PCM to analog -- which is was DSD was designed for (it's why DSD sample rates are integral multiples of CD's 44.1 sample rate) - basically amount to superstition.

 

[par4ken]

If purists were commonly obsessed with room acoustics and speaker accuracy, the two biggest impacts on home system sound quality, I would believe this to be a schism with a rationale (albeit weak*) behind it. But if the sound from a purist rig at a listening seat is highly colored by the speakers/room -- as is typically the case -- and that coloration can be more or less corrected by DSP, which one is offering higher 'sound quality' in the end?

Also, AVRs that have a DSD decoding chip in them always offer one or more 'purist' modes of playback that bypass DSP. To use these, purists should be using an HDMI connection to let the AVR receive the raw DSD bitstream from the disc/file player. Then there only is a single D/A conversion directly from DSD.


*worries about converting DSD to PCM to analog -- which is was DSD was designed for (it's why DSD sample rates are integral multiples of CD's 44.1 sample rate) - basically amount to superstition.

Purists are concerned more about a simple signal path, not coloured by DSP effects. Speaker room interaction is natural, and usually sounds good (usually) no reason to "correct" anything. I have built and tried out both graphic equalisers and a parametric equaliser, I also built a display to facilitate adjustment but found that the whole thing was a complete waste of time. All that is required are shelving type bass and treble controls, adjustable turnover is a nice feature so you can apply boost only at the frequency extremes if desired.

Higher sound quality is what you get subjectively in the end. Measurements are a useful tool but your ears are the ultimate judge.

Scully use your ears for a change instead of your DSP driven dogma!

In most cases HDMI should be reserved for video! HDMI is only becoming necessary now as more and more surround channels are added when four were just fine!

 

[J. PUPSTER]

Alright, I wasn't getting any help on this, so I re-ripped my DVD-A MLP 24x96 for EJ GBYBR, and yes all the channels are level matched at the peak 0.50 db for some reason; so not sure why the variance with the Fronts and the rest of the channels of ~1db for the SACD version?

Could it have something to do with the difference between MLP 24x96 and DSD 24x88.2?

 

[LuvMyQuad]

IMO a purist would eschew all DSP and treat the room itself using bass traps, absorptive panels, etc

One cant compare the analog EQ tools from the 70s/80s to the modern digital measurement packages you can get now. They are just too so different. Nothing.... I repeat nothing i have ever come across produces as much overall improvement in my system than using room correction. Its arguable better than upgrading speakers.

 

[himey]

That is ridiculous! On several occasions now my Oppo was playing SACD as PCM. I very quickly heard the difference, stopped the playback a switched back to DSD. The difference is very audible!!!! I didn't know it was set to PCM for sure until I looked at the display and sure enough. Sorry but I trust my ears!

Differences doesn't mean one is bad.
They are going to sound different because of dsp.
Are you saying all conversions to PCM with the Oppo makes it sound bad?
I convert to PCM using Saracon, and the resulting files are far from bad.
I trust my ears and my setup. I know what sounds bad.
Using terms like blows it away seems like you mean the resulting conversion sounds bad, which I highly doubt.

 

[LuvMyQuad]

Alright, I wasn't getting any help on this, so I re-ripped my DVD-A MLP 24x96 for EJ GBYBR, and yes all the channels are level matched at the peak 0.50 db for some reason; so not sure why the variance with the Fronts and the rest of the channels of ~1db for the SACD version?
Could it have something to do with the difference between MLP 24x96 and DSD 24x88.2?

Sorry Pup, I only have the SACD of GBYBR.

Dismissing the difference in the fronts for a moment, all else is within a fraction of a dB and on average about 6dB lower in volume on the SACD to PCM conversion, which is the same value noted at ther start of this thread.

 

[LuvMyQuad]

I convert to PCM using Saracon, and the resulting files are far from bad.

how do you go about that. is it a different plug in, or a different method all together?

 

[J. PUPSTER]

Sorry Pup, I only have the SACD of GBYBR.

Dismissing the difference in the fronts for a moment, all else is within a fraction of a dB and on average about 6dB lower in volume on the SACD to PCM conversion, which is the same value noted at ther start of this thread.

Maybe you just set your FB2K a conservative +4db overall as a hedge against some starting out higher than others?

 

[par4ken]

Differences doesn't mean one is bad.
They are going to sound different because of dsp.
Are you saying all conversions to PCM with the Oppo makes it sound bad?
I convert to PCM using Saracon, and the resulting files are far from bad.
I trust my ears and my setup. I know what sounds bad.
Using terms like blows it away seems like you mean the resulting conversion sounds bad, which I highly doubt.

I never said that anything sounded bad but I did say that native DSD did sound much much better and yes "IMHO native DSD blows away SACD converted to PCM" at least via my two Oppo's.

I might have to look into Saracon as well as other conversion tools as well. As I mentioned earlier even Foobar has many different settings, finding the best one might make all the difference. It is nice to be able to extract to PCM so that the mix can be altered, and for compatibility.
Edit: Did a search for "Saracon" priced at $2500 CAD totally out of my budget, I would hope that it works as claimed!

I've made CD's in the past from DSD files converted to PCM (by Foobar), they sounded better than commercially produced CD's but that's mainly due to the loudness war! Many recently produced CD's do sound truly bad!

I used to convert my SACD's to flac and never noticed a difference in quality but my old players all converted DSD to PCM. It was only with the Oppo BDP-103, playing DSD natively did I hear the difference! That is what I continue to "preach" about!

 

[par4ken]

Alright, I wasn't getting any help on this, so I re-ripped my DVD-A MLP 24x96 for EJ GBYBR, and yes all the channels are level matched at the peak 0.50 db for some reason; so not sure why the variance with the Fronts and the rest of the channels of ~1db for the SACD version?

Could it have something to do with the difference between MLP 24x96 and DSD 24x88.2?

View attachment 74544

I think that the 1dB difference is mostly insignificant.

 

[Kal Rubinson]

None!

But, logically, you cannot know that.