Listen and choose the 8th generation digital copy part Two

[esldude]

A chance to hear what multiple copies of DA and AD conversion do to a music file. 

 

This is part two of my other poll.  Might be best to keep most discussion in the first part.  If you missed it here is the link to part 1 below. 

 

I’m making this a poll, but you don’t have to vote. One of these files has been thru 8 generations of copying. So it is a chance to hear what damage is done with multiple digital copies of a 44.1 khz 16 bit file. Would be nice to hear from you if you listen to these to get your thoughts.

 

The two linked zip files will open into three wav files. One is labeled a Reference, one is A and one is B. Either A or B is the same as the Reference file. Please choose in the poll the one which is a copy. The copy has been thru 8 generations of DA to AD conversion. So it should be the odd one out on sound quality.

 

Each file is a 30 second snippet of a full song.

 

Your choices are:

A is a copy.

B is a copy.

They all sound the same to me.

They all sound different to me.

Skipped (meaning you skipped listening to this file)

 

https://www.dropbox.com/s/v60bqlqdfbcktf4/Jennifer Warnes.zip?dl=0

 

https://www.dropbox.com/s/1vnoquxjisu70zd/Bob Marley.zip?dl=0

 

[esldude]

38 minutes ago, fas42 said:

Found this thread before the other, and went for the Bob Marley - good choice, made it easy. The copy is all round duller, has lost sparkle, life and texture - has the listening to a transistor radio quality.

And pray tell which are you calling the copy.  Is it A or perhaps you chose B?

[esldude]

2 hours ago, sandyk said:

Dennis

 Which track and album is the Jennifer Warnes snippet from ?

 

Alex

The Famous Blue Raincoat

Track #2 Bird on a Wire. 

[esldude]

4 minutes ago, STC said:

Only 2 votes?  @esldude, did you mislable the tracks by any chance?

I double checked everything.  I don't think I did any mis-labeling.  PM me if you see something that indicates I did and I'll check on it. 

[esldude]

Just now, STC said:

 

Maybe not. Just started to listen to the other thread and I am judging on preferred SQ. Again - one of the copy sounded better than the original. That's not the purpose of this poll.

Would be interesting to know which one that was.  The purpose is to see if people can hear the difference.  So initially a difference is all the poll concerns.  

 

I'm not claiming any scientific basis for such a poll.  I'd hope it might be interesting for people to hear the differences without knowing which are which.  I think some would be very surprised how close such a copy sounds.  Not that it is necessarily indistinguishable.  I thought voting in the poll might garner a little more engagement with the topic.  As the usual when I've done listening polls, people seem reluctant to vote in them.  I didn't even make it a public poll.  So if no one leaves a comment they can vote anonymously and no one will know if they are wrong or right. 

 

And I don't think Alex ever votes in my polls. 

[esldude]

6 minutes ago, sandyk said:

 

Seems that you care enough not to use the IGNORE option ?

 

 Please enlighten me as to how 2 bit perfect images can consistently look different when using a non calibrated display, and FUTHERMORE, the preferred image never changes either, despite them being presented in a different order.

HOLD ON!

 

Please let us not get the bit perfect isn't always the same sound or image thing going.  Not the topic of this thread.  Please start a thread about that if you wish.  I'll even take part. 

 

These copies aren't digital copies they are copies done digitally via DA to AD conversion.   It is reasonable to think such after 8 generations would be audible (and it is).  But it isn't as easy as some might think if they've not heard it.  

[esldude]

21 minutes ago, One and a half said:

The Jennifer Warnes samples were downloaded and compared. The track is 'Bird on a Wire' from the 'Famous Blue Raincoat: The songs of Leonard Cohen' album. I play this album reasonably regularly and am very familiar with how it sounds, in fact, quite amazing.

 

Downloaded sample files are rubbish. Roon reports DR 2 compared to the ripped local version of DR9 (I think it's the anniversary CD in the library). The dynamics are lost and all three, A, B and reference are horrible. The B version has a focused centre compared to A of the 10s I could stand listening. That's what I noticed right away, by the time the reference was played, there was something really wrong and couldn't listen any more.

 

Played the ripped CD and that's more like the memory I have of the track, full of life, vocals front and centre and music spread from left to right, just classic. 

 

Have not answered the poll, the sources supplied are of insufficient quality to determine anything.

That is amazing. 

http://dr.loudness-war.info/album/list?artist=Jennifer+warnes&album=

 

According to this the Anniversary version has the lowest DR.  And the track I used off of the 1986 version I have is DR16.  I know it is bit for bit identical to the track on my CD.  And the software will use those bits for calculating DR.  So how does Roon come up with DR2 on any of those?

 

EDIT to add: Roon uses the R128 method so it isn't directly comparable to the DR database.  In any case, I don't see a DR 2 unless it has something to do with it being a restricted snippet of the song.  It is the same data as exists on the old fairly wide range CD. 

[esldude]

Here is a screenshot where I lined up the section I used for the reference. Below it is the entire track from the CD.  I inverted the excerpt.  you can see where I've highlighted a combined FFT for that section and it shows like -2000 something db which is what Audacity shows when there is nothing. 

 

Here is another screenshot where I've mixed the two together.  I've opened the contrast window and for the highlighted section under volume it shows 'zero'.  It will only show this if there is nothing except zero value samples in the selected area.  Sounds like someone looked at Roon's DR and decided what they would hear.  As to why Roon shows this I don't know. 

 

[esldude]

Okay, I expected some adult reading  and context comprehension.  I could have worded the headline title differently I suppose.  If I had used all the proper exact most precise king's English, I think we all know Alex still wouldn't have listened and voted.  While not the case here, what if I had used all the same file to see if people heard differences or instead selected I don't hear a difference?  

 

My snippet is an absolute bit for bit identical copy from the original CD which is the CD which caused this recording to get a reputation as an excellent bit of music and excellent recording.  Now to some it sounds like a transistor radio.  Never heard an old AM transistor radio (the connotation I gathered) with 20hz -20 khz response.  I've never been to Australia unfortunately.   One of the files has indeed been thru 8 generations of copying via a DAC feeding an ADC.  I've listed the names of the gear.  I very, very slightly tweaked volume on the copies to match the original.  All recording was done 24 bit, and I took the final 8th generation copy and saved it into dithered 16 bit.  All of which if anything would make the 8th gen copy much easier to hear as different. 

 

Alex has a version of the recording which sure appears to have been through one sloppy DA to AD conversion.  The level is lower by nearly 3.5 db. Transients are bit blunted.  And there are a couple other tell-tale signs.  He doesn't know the provenance of it, but it sounds far better he says than my exact copy from the original CD.  I suggest everyone get a copy from him and see what you think. 

 

Finally these files are discernibly different.  It is very difficult.  Which was the main point of this.  An 8th generation copy with measurable amounts of degradation is still good enough after going thru a $380 DAC that you have no easy time hearing a difference.  

[esldude]

59 minutes ago, sandyk said:

 Really ?

 Perhaps you are happy with .mp3 too ?

 Your posted personal conclusions have just taken out the Audiophile from Audiophile style and made a mockery of this forum and what it stands for, as can be seen in other areas of the forum.

 You aren't just talking about digital copying here, you are talking about numerous D to A and A to D conversions,

and highlighting just how pathetic some cheap ($380) DACs can be if you have difficulty hearing the differences .

So let me get this right.  I use a cheap DAC to 'defile' 8 trips thru it.  Then it vs the original file is harder to discern than if I used a tremendously good DAC compared to the original?  You'll have to explain that one to me.  Your conclusions seem backwards.  

 

So yes I'm giving an example of numerous recordings and conversions with inexpensive gear.  Seems it would be a piece of cake to hear if the gear is so poor as you imply.  

 

And no I've not seen emails as I was out on an errand.  I'll look. 

[esldude]

@sandyk Have you ever heard any song I've posted that was a good recording?  I seem to recall you claim they all sound very poor and substandard.  Seems unlikely I somehow always have the worst version of everything.  And did you listen to any of the other 4 songs in this thread?  Are they all so poor quality you can't make heads or tails of them?

[esldude]

Looking at info about versions of Famous Blue Raincoat I knew there were many illustrious musicians that had some part in it.  I didn't know that some of the guitar work on it was by Stevie Ray Vaughn.  If my files were less lacklustre I probably would have heard Stevie's licks on the songs.  

[esldude]

5 minutes ago, Ralf11 said:

I suspect what is happening is that the change in coriolis force is making them sound lacklustre to some people.

So you are going with the coriolis force being the opposite direction in the down under.  I like it.  

 

Let me ask your opinion about something.  Should I turn my listening room orientation to an east-west direction to prevent a channel imbalance from north south orientations?  And should I face east or west? I'm in the northern hemisphere.  

[esldude]

12 minutes ago, Ralf11 said:

You must align your Axis of Auditory Integration with the magnetic field

 

 

unfortunately, it is shifting...

Yes I read about the recent rapid shift.  Fortunately there is a place to keep up to date. 

 

http://www.magnetic-declination.com/

 

I'm 3.5 degrees negative.  30 years ago when I was an active private pilot before GPS my location was on a 0 degree line. So do I face west, and plus 3.5 degrees south?

[esldude]

5 minutes ago, sandyk said:

 How about you actually read what I have posted ?

 I stated that I could still  hear clear differences with the Jennifer Warnes tracks , but I refuse to vote in Dennis's feeble attempts to prove that cheap DACs etc. are revealing enough, and that we don't need the more expensive equipment available from the Advertiser's and Sponsors of this forum, which wouldn't exist without their support.

There has also been a pretty pathetic voting response to this thread from the large volume of readers and members.

 This speaks volumes. The vast number of participating members are clearly more interested in obtaining further improvements from their Audio gear as evidenced in Rajiv's massive thread.

It has also become quite clear that some members resent the name change of the forum, and the implied accent on Audiophile products which several from the Anti Audiophile brigade appear to believe are Snake oil products..

 I have better to do than waste my time discussing  this further with you.

So you are protesting the idea cheap dacs show differences by not seeing if they do by not listening? Or are you protesting my files being so lousy no difference can be heard? I'm confused except I knew you wouldn't vote. Just confused about what the reason is this time. Quite the comedian.

[esldude]

Since the poll is closed here are the answers for each file. 

 

1 is Ry Cooder and V.M. Bhatt B is the copy.

2 is Phil Woods A is the copy

3 is Cowboy Junkies B is the copy

4 is Jennifer Warnes B is the copy

5 is Bob Marley A is the copy.

 

1 B is the Copy

2 A is the Copy

3 B is the Copy

4 B is the copy

5 A is the copy

 

 

So 50:50 split on the first two and 2, 4 split on the third.  

 

I can ABX them all.  It isn't easy, but it is doable.  I can't do the 4th gen files or even the 6th gen files.  I can the 8th gen files.  

[esldude]

1 hour ago, fas42 said:

 

Doing a bit of analysis of what's going on, DiffMaker type of thing, it can be clearly seen that the higher frequencies, above 5kHz, are distorting significantly, after the 8th generation copy; and, particularly interesting anomalies in the left channel, not far down from the peak levels for that range of frequencies. Is it the music signal, or is it a greater weakness in that channel?

 

What's clear is that the nominal, measured precision of the converters is not realised, in practice ...

I've not noticed that distortion.  I'll take a look at it.   Nothing ever approached clipping anywhere along the way.  And I'm not seeing it.  Can you point to which one and maybe what part of the track you see this distortion?

 

BTW, you should try pkane2001's new software.  Deltawave.  Does what Diffmaker did only better, with more useful features and works reliably without blowing up.  It is still in beta and being improved rapidly.  

 

https://deltaw.org/#

 

[esldude]

1 hour ago, fas42 said:

Thanks! I'll check out DeltaWave, to see if I can use it to see what I was able to obtain, just using manual manipulation.

 

No clipping, rather the alteration of the high frequency content - an example just looking at the 5k - 20kHz content,

 

In that track Frank, there is a difference in higher frequencies in the two channels.  The left has some spikes in the FFT around 5150 hz and 7979 hz in the left channel which are missing in the right.  You can hear it if you listen to each channel in mono.  

 

So nulled out the original vs 8th gen copy is in the mid 40 db range.  So the residual of the left channel had more energy there, and left a greater residual in the left channel than in the right.  This even though the level of the null is close for both channels.  

 

Here is the left channel in the original file.

 

And the right channel, both taken a couple seconds around the 15 second mark.