Bowie – "Heroes": Blind listening test

It turned out it was just displayed differently in another audio editor.Same mastering and a waste of 24 bits

 

Oh, thanks – I didn't know it turned out to be a mistake in the display. Anyway – it's hardly brick-walled. But on second thoughts, I really really don't want to start that old conversation up again.

I think this blind test will be interesting.

 

I kind of want to hear the start of Beauty and the Beast because that was when I first noticed subtle differences between the WG, the Ryko and the 2017 and I don't mean the fuller bass.

 

Here are my own thoughts on this, after having further randomized the samples I listen to, so that I don't know which sample is which. With a pair of DT770s and various DACs.

A: too bright and harsh, right from the first cymbal sound on Beauty & the Beast.
B: a bit too fat for a 70s record, but I could live with it.
C: nicely warm, for example on the handclaps to The Secret Life of Arabia. But the same song sounds a bit muffled, compared to the other samples.
D: manages to sound artificial and manipulated even on a soft track like Moss Garden. The handclaps on Arabia sound painful, now.
E: same as A, with maybe a little less bass.
F: a bit brighter than C, but still a good bass, sounds quite natural.

My ideal master would be somewhere between C and F, that is: the warm sound of C, without its (small) muffled effect. B comes after this, and everything else sounds really unpleasing.

On my previous DAC, I'd have chosen F, but my new one (just from this week) is flatter so I think I prefer C, especially since I find this album a bit fatiguing in the long run. I think a better master than C could be achieved, though.

So, in this order: C, F and B.

 

I threw my vote away as I spent too much time on Beauty using cheap AKG headphones at work and couldn't decide on anything. On my hifi at home, I decided C was good but a bit muddy, and I thought I preferred F. But what really convinced me overall was the final track Arabia where I thought I could hear more difference in the sound – ending up thinking less of C and more of F.

What i can't work out is how low the volume is overall. To match the normal WG or Ryko levels I need to add 5 to my hifi volume (from 30 up to 35). And I also kept wondering which was the 2017 as none of them sounded like it. And that led me to think that somehow the reduction in gain had tampered with the sound quality. Probably I'm wrong but there it is.

It's been interesting as I find read back to agree with the general feeling about the qualities of the samples. Thank-you Tsomi. I'm going to be embarrassed when I find out that 2017 remaster is sample D.

 

Do you mean the Beauty and the Beast sample should start a bit earlier? Or that it's fine as it is? I couldn't make it start from its real first notes, though, since one of the masters is missing some opening notes, so it would have been really easy to identify it and have bias for/against it.

It mostly matches the most dynamic available master, while avoiding clipping. So, if you're used to the volume of the EMI-1999s (or some Parlophones), there's going to be something like a 6 dB difference for those, yes. Unless we want to introduce clipping, the modern mastering will almost always lose some dBs from its usual volume, because of this. I don't think it should really impact its sound quality, though.

However, I really don't understand this feeling if you're comparing this to your normal RCA WG. Up is the WG sample (with ReplayGain pre-applied to the file), down is the straight rip of the track from EAC, with absolutely zero processing. If I compare the same segments, there's only a 0.3 dB difference (Ashes to Ashes).

That shouldn't happen, but it's still possible, unfortunately. I still have the "raw" samples, where I've only cut segments from the straight EAC rip with zero processing (no dithering, no level adjusting, nothing, just cut, so they're the very same bits). If you'd like to help, and if you manage to find a way for them to have the same perceived loudness, without the bad processing that you describe, then yes, please help me. And if people really hear an improvement between my samples and your samples, then there is something wrong with my method, and I'd need to fix this ASAP.

I need someone to work with me if you really think there's an issue here.

Thank you for the kind words!

 

Tsomi,
thank-you for the explanations. I understand what you have done. I think you know about these things better than. I do not know if there is an issue it was just a thought.

So, I understand why you do not use the start to B&Beast.

With regards volume. Here is your sample 1 (1) against my WG RCA Beauty and The Beast lossless ripped (2). Your sample seems to be much quieter.





With regards the 2017 remaster – I do wonder whether if an audio file has the gain reduced from 1 to 0.5 whether it effects the audio quality. Surely a signal represented at the lower end of the bit-depth (0 - 0.5) has less bits used to represent information than one that takes the whole range from 0 - 1.0? Or am I wrong?

Having said all that I know that the 2017 Heroes has quite low DR (8-10), so I am not expecting it to come out on top. Only finding out which sample is which will enable to me to confirm my thoughts about any relationship between gain and audio quality.

I'll wait for the identities to be revealed.

 

Well, first, I'm sorry, I mixed up the Scary Monsters and "Heroes" threads while replying here, so my screenshot above was not comparing the tracks you were talking about here, sorry.

You're right, there's a significant difference in volume between an original RCA WG and the WG sample I'm including. It looks like this is because foobar chose the RCA JPN as a reference, and its volume level is even lower than the RCA WG. Moreover, foobar keeps a lot of headroom by default it seems, and something like 3 or 4 more dB could have been kept in all the samples without introducing any clipping.

(Your screenshot compares a segment with a full song which makes the problem seem even bigger, though, and the sample you chose is not the WG, but that's not a big deal.)

I don't think it had much impact on the way the songs were being heard (since all the samples were a bit too low), except maybe for people who wanted a bit more volume to do better comparisons? I'm sorry if this is the case, and I'll make sure the volume is not lowered too much without any good reason for the next tests, thanks (I remember doing this for the Young Americans test, though).

I don't have enough scientific knowledge in this... But yeah, I think it does have an impact, but is it that big, compared to EQ/denoiser/compression differences between masters? The samples need to have a similar perceived volume, otherwise there's the "louder is better" bias. I can't make the older masters as loud as the latest ones, because they'd clip a lot and that would introduce a lot of distorsion. They have to meet somewhere, the question is: where? To an average value? I thought the algorithms in ReplayGain would just pick the best value, from an audio quality perspective.

You can PM me if you want the results, since you've already voted.

 

I found it really hard listening to them and deciding which I preferred. I did it through some kind of instinct. To be honest I'm not sure what I am listening out for. I know that when I find a master on my hifi it just 'clicks' on different levels. Whilst I did choose one or two and rejected others, none of these especially clicked for me. So, I guess I'm saying that listening to a lot of samples of a lot of songs is a strange kind of activity that needs a specialised state of mind and affects the outcome.

I can't help thinking that those people above who identified (or think they identified) the sources of the samples are using some pre-test memory of what each master sounds like. ie. They are bringing knowledge to the test, that others do not have. Likewise I guess that some people are 'experts' at these kind of listening tests. So, what I am saying is that a test like this is not really objective because of the varying skills of the listeners but also the pre-knowledge of some listeners will effectively mean that the test is not really 'blind'.

The next thing that confuses me is why do I enjoy the compressed and souped-up 2017 remaster so much - in comparison to my WG copy. But then I find that in the test I am listening in a different way that makes me choose another master.

Do the tests seems to have an effect on perception that change the results?

All enjoyable and informative, nonetheless!

PS. Tsomi – I altered the gain on the 2017 lossless file (around -8 db) and it does sound exactly like your version of the same. So, again a perceptual issue on my part.

 

These tests are indicating to me that I was right to hang on to my (Spoiler) as that's what I have a tendency to pick! I do love my RCA WG's though for their overall familiarity, lack of tape damage and so on.

 

Yes, that's possible, but there's no way to remove this from their heads . But then, are these preconceptions confirmed by the test? Big connoisseurs of the different Bowie masters can add very informative opinions to the discussion (and there are some RCA fans, but also some Ryko fans, some EMI 1999 fans, some Parlophone fans), and I'm really happy when I see some people who are not very familiar with the material (or with the different masters) come and vote here. For the same reason, I've voted in some Pink Floyd blind tests; I'm not very familiar with the material (not a really big fan) but I thought that it's what made my vote interesting among others.

There's a surprise in the current TMWSTW votes for example; I think some people will be quite surprised by what they voted for.

Yes, that's true. I know karmaman now waits before submitting his own feelings in these tests, so that less votes would be influenced by his own opinion. I think a lot of people do read the existing opinions before voting themselves, so here's another reason why it's not really blind either. I really wish people would vote before reading anything, but well... the output is still interesting.

I think, unconsciously at least, a louder master will often appear better. I quite like the "Heroes" 2017 Parlophone remaster myself, and I think I could live with it if it didn't have some really big mastering errors. The dynamics have been compromised, but I almost can't hear this problem on this particular remaster, when it's leveled with the other masters (I'm 25 so my ears are quite used to modern mastering, unfortunately). I prefer the EQ of another master when they're compared with attention, but the difference wasn't really big for this particular album, for me (I had much bigger problems with the Parlophone Young Americans sounding really muffled).

I don't think these tests show anything definitive. It can reveal some things, some tendencies, some misconceptions. But it's just a hint. You can try to do more comparisons on your own, between your previous favourite master, and the one you've voted for, here. With a full listen, your own adjustments, or whatever. There's absolutely nothing wrong in continuing to prefer listening to your Parlophone if you prefer it, really

I'm done voting in the TMWSTW test, but I still don't know what is my go-to copy, after doing this. I might pick the Japanese LP recommended by karmaman, because it sounded like a good balance between the different masters I've voted for. For Young Americans, the blind test helped me a lot and now I'm sure the RCA JPN for US is my favourite digital master. Some people here said that some tests were much easier than others. It's going to be an interesting journey, in the end.

Thank you very much! No problem.

 

I’ve been doing my own tests and now hear some kind of channel wavering or dropout on Sons of the Silent Age 2017 CD, very noticeable on headphones. Just before the final section of the song. It’s not on my WG RCA CD. Haven’t checked any old vinyl yet.

 

Yes I hear a slight dip in the right chanel just after 'go to sleep one day'

 

Well, um, er, I made sure to include an EMI '99 of "Lodger" on my personal "bonus discs" homemade jewel box to go with box3. (I made this decision before acquiring a 96/24 vinyl rip, which will likely be my go-to forever. I had read what seemed to be dire warnings that the best "Lodger" anyone could hope for would be muddy at best, but that's not really true.)

To me the problem with the '99 EMI Lodger was deeper than the harshness or brightness. The process they used - whatever it was - to my ears seemed to have squeezed some actual individual instruments and sounds out of existence. But OTOH, maybe it will be more amenable to home-brewed EQ experiments. I dunno. Maybe when I'm 64 it will be the only one I can still hear. I don't have to lose sleep over it, I just have to make 350MB of room on a disc somewhere. No biggie.

To make a long story short, I know this '99 Lodger reference was probably meant as a joke, but I was just chiming in with something to be said for it.

 

Some tracks sounds worse on 99 remaster, like Red Sails. But it brings DJ, Repetition and some other tracks to life from that murky swamp of a Visconty mix. I keep it as a bonus too.

 

A - weak bass, cold on Heroes
B - Strong bass, perhaps the best Beauty, with a nice oomph in the kick drum, piano/vocals on Arabia clearer than C
C - Warm with good bass, muddier on Beauty, drums sound a bit distant, thuddier snare on Heroes
D - good bass thump on Blackout, brighter snare than B on Heroes, but the almost non-existant left channel on Arabia kills this one
E - weak bass, Heroes too bright
F - not bad, pretty good all around

This one was much tougher for me than TMWSTW, less variation in the vocal sound. I think I'd give a slight nod to B, with C in second

 

Listening off a Macbook with Audio Technia ATH AD700, via XLD, so nothing hi-end. Like most it was easier to spot what I didn't like: A, D & E and harder to pin down which was best from B, C & F. Listening to Moss Garden, particularly the gong? sound in the last part, it seemed F captured it better, but at other times the vocals were smoother on B & C than F, obviously this might change playing it through speakers, I could probably live with any of these 3.

 

This was a much harder choice for me to pick compared to the one for The Man Who Sold the World. My favorite one out of all of them is F. It was the most balanced one to me, all the tracks had a good amount of top end with sacrificing the bass. C has a very warm sound, which can be a plus for a song like "Heroes", but can be a fault for others like Beauty and the Beast, where it sounds very muddy compared to the other samples. B has a good bass, but can be a bit too much for me. A doesn't have enough bass, and E is pretty close to the sound of A. Then lastly, D is clearly the worst. Everything just sounds totally off, too bright and trebly.

 

I noticed a typo on my post. I meant to F has a good amount of top end without sacrificing he bass.

 

Well, the box is nearly a month old, and, pending the re-release of "Heroes", I finally burned the last of my supplemental "bonus material", and exhausted my meager graphic talents doing "liners" (I'm too lazy to make anything but spines).

It's a 6-jewelbox box set for me.

1) 192/24s of the studio albums, from Parlophone, on two DVD-Rs
2) 192/24s of the two Stages from Parlophone plus a rip of the 24/48 LPCM 2005 DVD-A (two DVD-Rs)
3) The "RCA" set: illicitly obtained rips of four RCA Vinyl (96/24) and 5 RCA CDs (two DVD-Rs)
4) The "EMI" set: 4x EMI's 2009 SHM-CDs (since "Lodger" was already in for the long haul, I figured "Everybody into the pool!"), the 24/48 Stage rip (again), and the "Scary Monsters" DSD files. (two DVD-Rs)
5) What I call the "Remix" jewel box: Low live in 2002 (concert boot), "Scary Monsters Chronicles" (outtakes boot), and a blank slot for whatever they decide to put out to fix the "Heroes" album whenever they decide to do it. (1 disc in a dual-disc box, so far.)
6) I put the "Re:Call3" disc in a 1-disc jewelbox because I detest mini-LP sleeves.

 

When do we get to find out? The waiting is driving me mad! PM?

 

For TMWSTW, it's going to be "soon". I'm letting a bit more time for Heroes and Scary Monsters, as long as new votes pop up from time to time. PM me if you want earlier results, thanks.

 

Tsomi, are you going to do a test for Low?

 

I'd like to do, at least, the full 1969-1980 discography. If time and motivation permit it, I'd like to do as much as possible from the full list of studio albums.

Diamond Dogs is next, probably for this week or the next one. Low/Lodger are not far away. I'd like to do Space Oddity and Hunky Dory as well (I'm quite intrigued by the possible results of the Space Oddity test). The Glam rock stuff will come later, because there are so many available masters.

(I'd like to do some other artists as well, but time is finite )

 

F for me.

A is bright, almost screechy and bass shy.
B has more bass, is a more balanced offer, natural vocals, second best for me.
C is a bit too bassy for me.
D has more bass than A or E, is also trebly, has a wonky midrange and sounds (despite it's brightness) dull and lifeless. The worst of the bunch for me.
E is similar to A for me.
F is not the last word in details, but it's a very balanced offer