Dead Can Dance - Aion - Radharc

[manisandher]

Here are three different 16/44.1 files:

 

- one is a recording I've made from vinyl (180g boxed set)

- one is an SACD rip (using a PS3, then converted to PCM using Audiogate)

- one is a CD rip

 

Can you guess which is which?

 

Mani

 

[wkhanna]

I will not guess, but can hope for an outcome favorable to a certain format.

 

And I can wish for that Tascam unit for use when ripping my vinyl.

 

Nice rig, looking forward the final tally.

[AudGuy]

At a guess, I would suggest that:

 

file 1 is the CD rip

file 2 is the vinyl rip, and

file 3 is the SACD rip.

 

BUT, I am guessing and could be completely wrong also ..... ????

[89reksal]

I'll go with:

 

file 1 SACD

file 2 CD

file 3 vinyl

[Julf]

Can you guess which is which?

 

Could you post the spectrum plots too?

 

Independent of what the real order is, I think what can be proven is that any of the popular formats in use today are more than capable of reproducing the normal dynamic range of the recordings, but for some reason artists, record labels, producers or mastering engineers (pick your favourite target to blame) choose to compress material more for some formats and less for others - probably for commercial reasons.

[manisandher]

Well done AudGuy, spot on.

 

Could you post the spectrum plots too?

 

Here you go:

 

 

They look remarkably similar, don't they? And so they should - they're all derived from the same original 1990 digital master (which I strongly suspect was 16/44.1).

 

Apart from the difference in levels, the CD and SACD spectra are virtually identical (up to 22.05KHz - not shown, but above this the SACD shows increasing noise, as per all DSDs). The CD however is quieter low down, whereas the LP and SACD are remarkably similar.

 

There seems to be very little 'signal' above 15KHz. From 15KHz to 22KHz, the LP shows noise at around the -85dB level - the CD and SACD don't show this.

 

Independent of what the real order is, I think what can be proven is that any of the popular formats in use today are more than capable of reproducing the normal dynamic range of the recordings, but for some reason artists, record labels, producers or mastering engineers (pick your favourite target to blame) choose to compress material more for some formats and less for others - probably for commercial reasons.

 

I couldn't agree more.

 

Why the compression on the CD and then the low level? And why the compression on the SACD and not the LP?

 

FWIW, I prefer the sound of the LP to the 16/44.1 CD rip or the DSD64 SACD rip. It's just more engaging. But NONE of these graphs gives any clues as to the differences in sound that I'm hearing.

 

Mani.

[manisandher]

For completeness, here are two further spectra:

 

SACD DSD64 rip (with PS3) to PCM24/176.4

 

 

Vinyl to PCM24/176.4

 

 

Mani.

[Julf]

For completeness, here are two further spectra:

 

Thanks - they do, once again, show why a 48 kHz is more than enough for storing and distributing material sourced from vinyl or DSD.

[Jud]

Could you post the spectrum plots too?

 

Independent of what the real order is, I think what can be proven is that any of the popular formats in use today are more than capable of reproducing the normal dynamic range of the recordings, but for some reason artists, record labels, producers or mastering engineers (pick your favourite target to blame) choose to compress material more for some formats and less for others - probably for commercial reasons.

 

I wonder whether the usual higher DR ratings for vinyl are a result of compression applied in other formats or an artifact of the process/technology used to arrive at the DR figure.

[manisandher]

Thanks - they do, once again, show why a 48 kHz is more than enough for storing and distributing material sourced from vinyl or DSD.

 

Julf, I think there's a lot more to it than this, and I suspect a lot of it sits in the temporal domain, as opposed to the frequency domain. For example, my 24/176.4 recording from vinyl sounds way better than my 16/44.1 or 24/88.2 recordings from the same vinyl.

 

If anyone's really interested in hearing this for themselves, there are some links to these files here: Musings on digitizing vinyl. But there's just one caveat - I think you really need to use a non-oversampling DAC to hear the true differences, because the 24/176.4 was recorded using a true R2R ladder converter with a non-oversampling filter. And there aren't many 24/176.4-capable NOS DACs out there.

 

Mani.

[manisandher]

I wonder whether the usual higher DR ratings for vinyl are a result of compression applied in other formats or an artifact of the process/technology used to arrive at the DR figure.

 

Jud, that's why I included the waveforms. It's clear simply from looking at these that the vinyl has a higher DR.

 

Mani.

[Julf]

I wonder whether the usual higher DR ratings for vinyl are a result of compression applied in other formats or an artifact of the process/technology used to arrive at the DR figure.

 

Unless the reference levels of the recordings are matched, you can't compare the DR values. For digital data a reference level is easy to establish, as you know exactly how much "full scale" is. For vinyl there is no such absolute reference level.

[Julf]

Julf, I think there's a lot more to it than this, and I suspect a lot of it sits in the temporal domain, as opposed to the frequency domain. For example, my 24/176.4 recording from vinyl sounds way better than my 16/44.1 or 24/88.2 recordings from the same vinyl.

 

Mani, I am not saying the difference in input antialiasing filter steepness doesn't make a difference. I definitely think you need to record in higher sample rate, but then apply high-quality, off-line digital filtering (and only to the extent necessary to avoid aliasing). Once you have done that, 48 kHz is more than enough for storing and distributing the result.

[Julf]

I wonder whether the usual higher DR ratings for vinyl are a result of compression applied in other formats or an artifact of the process/technology used to arrive at the DR figure.

 

Well, it really doesn't matter, as the recordings probably are from different masters. Looking at the graphs, it is clear that the processing/compression is different, and has nothing per se to do with the actual media used.

 

I initially, mistakenly, thought the 3 recordings were all done by Mani, with full control of any processing. Re-reading the initial posting that clearly isn't the case. So the only conclusion we can draw is "recordings can be processed differently when mastered for different media".

[Jud]

Well, it really doesn't matter, as the recordings probably are from different masters. Looking at the graphs, it is clear that the processing/compression is different, and has nothing per se to do with the actual media used.

 

I initially, mistakenly, thought the 3 recordings were all done by Mani, with full control of any processing. Re-reading the initial posting that clearly isn't the case. So the only conclusion we can draw is "recordings can be processed differently when mastered for different media".

 

I was thinking about what seem to me to be the generally higher DR values for vinyl, even compared to CDs of the same vintage, at DR Database, and whether those differences and the differences for these files in particular might be attributed to some artifact of the tool and the way it measures DR. But with what Mani's said about their provenance and the appearance of the waveforms, I understand that's not the case (or at least not the primary reason for the DR difference) with these particular files at least.

[manisandher]

Looking at the graphs, it is clear that the processing/compression is different, and has nothing per se to do with the actual media used... the only conclusion we can draw is "recordings can be processed differently when mastered for different media".

 

Yep. As so many people are saying nowadays, the mastering is more important than the format.

 

However, my own belief is that oversampling in the ADC and/or DAC affects the sound negatively. In which case the format does play a part because you need to use at least 4fs to get away with a non-oversampling AA filter in the ADC.

 

Mani.

[Jud]

Julf, I think there's a lot more to it than this, and I suspect a lot of it sits in the temporal domain, as opposed to the frequency domain. For example, my 24/176.4 recording from vinyl sounds way better than my 16/44.1 or 24/88.2 recordings from the same vinyl.

 

If anyone's really interested in hearing this for themselves, there are some links to these files here: Musings on digitizing vinyl. But there's just one caveat - I think you really need to use a non-oversampling DAC to hear the true differences, because the 24/176.4 was recorded using a true R2R ladder converter with a non-oversampling filter. And there aren't many 24/176.4-capable NOS DACs out there.

 

Mani.

 

Well, I wonder if you really need an NOS DAC. I understand you might need a mighty fine DAC, and perhaps that eliminates any DAC which does its own oversampling rather than leaving sample rate conversion, if used, to the computer source. Did you do any listening using sample rate conversion in XXHE and compare it to feeding 176.4 resolution to the DAC? (I'll leave the subject of any other XXHE settings adjustments to the Phasure forums.)

[manisandher]

I was thinking about what seem to me to be the generally higher DR values for vinyl, even compared to CDs of the same vintage, at DR Database, and whether those differences and the differences for these files in particular might be attributed to some artifact of the tool and the way it measures DR. But with what Mani's said about their provenance and the appearance of the waveforms, I understand that's not the case (or at least not the primary reason for the DR difference) with these particular files at least.

 

It's a total mystery to me as to why the CD has such a poor DR. It's not the case of having been caught up in the loudness wars - I bought it in 1990 and it's not at all loud! The SACD and vinyl were released at about the same time as each other (2008ish), and I'm pretty certain have the same provenance (look at the spectra - they're so similar). So why a higher DR on the vinyl? I have no idea. Maybe my London Decca Reference cartridge is adding some extra umph? I don't know...

 

Mani.

[manisandher]

Did you do any listening using sample rate conversion in XXHE and compare it to feeding 176.4 resolution to the DAC?

 

No I haven't done this... yet.

 

What I did do was 'upsample' the three vinyl recordings (16/44.1, 24/88.2 and 24/176.4) up to 705.6 in XX when listening back, to create some sort of level playing field. Otherwise there would have been massive imaging when playing back the lower-res files. What's interesting is that the 16/44.1 and 24/88.2 recordings sound very similar in 'quality', the latter sounding a bit 'fuller' and more 'organic'. But the 24/176.4 sounds fundamentally different - 'airy' and 'delicate', with a real rhythmic drive.

 

I believe there's something fundamental going on in the ADC. Getting concrete information is neigh-on-impossible. What I do know is that the PMII works internally at 24/176.4 (or 192) and decimates down for lower SRs. But at 176.4, uses a non-oversampling AA filter. I'm pretty sure the lack of decimation is a big factor towards its better sound at this rate.

 

Mani.

[Julf]

It's a total mystery to me as to why the CD has such a poor DR. It's not the case of having been caught up in the loudness wars - I bought it in 1990 and it's not at all loud! The SACD and vinyl were released at about the same time as each other (2008ish), and I'm pretty certain have the same provenance (look at the spectra - they're so similar).

 

But if you look at the amplitude plots, they are definitely different, so even if they come from the same master, there clearly is a difference in the processing (including limiting or compressing) that has been applied. It is also strange that both the Cd and SACD versions, despite being limited/compressed to a very uniform max amplitude, have some sort of transient at the end that goes above the limiter level.

[manisandher]

It is also strange that both the Cd and SACD versions, despite being limited/compressed to a very uniform max amplitude, have some sort of transient at the end that goes above the limiter level.

 

Yeah, I wondered about this too.

 

Mani.

[Julf]

I was thinking about what seem to me to be the generally higher DR values for vinyl, even compared to CDs of the same vintage, at DR Database, and whether those differences and the differences for these files in particular might be attributed to some artifact of the tool and the way it measures DR. But with what Mani's said about their provenance and the appearance of the waveforms, I understand that's not the case (or at least not the primary reason for the DR difference) with these particular files at least.

 

Well, it could still be the case even with these recordings. As we can see from the graphs, they are not at consistent levels, and the vinyl copy is most probably not normalized to a reference level - so the DR measurements can be way off.

[manisandher]

Hey Julf, I agree that the absolute DR values may not be accurate, but the differences in DR values are easy to see. If you told me that the CD rip was 9, I could pretty much guess that the SACD would be slightly higher and the vinyl much higher, just from looking at the waveforms. No normalization required.

 

What's interesting is that the CD rip sounds louder to me than the 16/44.1 recording, even though its rms and peak outputs are lower. I definitely would not have guessed the relative DRs just from listening.

 

Mani.

[Julf]

Hey Julf, I agree that the absolute DR values may not be accurate, but the differences in DR values are easy to see.

 

Agree. So in that sense they have to be regarded as different recordings - they have been processed differently, and the differences are not caused by the media as such.

[Jud]

No I haven't done this... yet.

 

What I did do was 'upsample' the three vinyl recordings (16/44.1, 24/88.2 and 24/176.4) up to 705.6 in XX when listening back, to create some sort of level playing field. Otherwise there would have been massive imaging when playing back the lower-res files. What's interesting is that the 16/44.1 and 24/88.2 recordings sound very similar in 'quality', the latter sounding a bit 'fuller' and more 'organic'. But the 24/176.4 sounds fundamentally different - 'airy' and 'delicate', with a real rhythmic drive.

 

I believe there's something fundamental going on in the ADC. Getting concrete information is neigh-on-impossible. What I do know is that the PMII works internally at 24/176.4 (or 192) and decimates down for lower SRs. But at 176.4, uses a non-oversampling AA filter. I'm pretty sure the lack of decimation is a big factor towards its better sound at this rate.

 

Mani.

 

What I'd be curious to learn is what you think of the 24/176.4 with no oversampling in XXHE vs. the same recording oversampled/interpolated to 705.6 - in other words, whether your preference is for one less sample rate conversion, or higher resolution obtained through post-processing rather than in the original recording.