Post by: azuolas on November 19, 2005, 11:02:10 PM
I uploaded the clips that were used during the blind ABX test yesterday on the CRC FTP site. For the sake of simplicity I would recommend everybody to post short and simple answers and defer discussions to other threads.
e.g. "Number x has a loss of low octave plain as a day" or "I can not hear any loss in any of them" or whatever.
All files are 24 bit 96 KHz WAV. These are 15 second clips.
I labeled all configurations randomly as 1-9. Unless there are objections I will post the answers to what these numbers are on November 23rd at 11:00 PM CST.
To download:
go to: http://www.chicagorecording.com/
Click on icon CLIENT FTP (3rd from the right)
username: chicagotest
password: prosoundweb
Then click on folder named "15 second clips 2496"
Download all 9 and test away
Azuolas
Post by: minister on November 19, 2005, 11:33:59 PM
tried this link in safari and IE, didn't work.
launcghed TRANSMIT and tried to do it that way. no luck...
Post by: CaptainHook on November 19, 2005, 11:43:11 PM
I.e http://www.chicagorecording.com/
Post by: minister on November 20, 2005, 12:01:09 AM
Post by: groucho on November 20, 2005, 01:42:48 AM
Mixerman has got to get credit for the longest-running practical joke in the history of the world. Orson Wells ain't got nothing on him.
What's funniest is he'll find a way to continue it even given the evidence. I can't wait to see how he does it. The guy's a genius.
Enjoying every minute,
Chris
Post by: azuolas on November 20, 2005, 01:50:03 AM
| groucho wrote on Sun, 20 November 2005 00:42 |
So... one of these files is supposed to be missing an octave of low end? |
If mixerman's claim is correct 6 of 9 of these files should have the bottom octave missing.
Post by: RKrizman on November 20, 2005, 02:02:35 AM
| azuolas wrote on Sun, 20 November 2005 01:50 | ||
If mixerman's claim is correct 6 of 9 of these files should have the bottom octave missing. |
You mean 0 - 20 hz?
-R
Post by: azuolas on November 20, 2005, 02:32:42 AM
| RKrizman wrote on Sun, 20 November 2005 01:02 | ||||
You mean 0 - 20 hz? -R |
No I mean whatever the mixerman's claim was. 50Hz or 60Hz? Whoever was a part of pretest discussion please jump in and correct me if I am wrong.
Post by: RKrizman on November 20, 2005, 02:41:41 AM
| azuolas wrote on Sun, 20 November 2005 02:32 | ||||||
No I mean whatever the mixerman's claim was. 50Hz or 60Hz? Whoever was a part of pretest discussion please jump in and correct me if I am wrong. |
You're quite right. I was making a joke ( I mean, what is the bottom octave anyway?) The claim was that there was a "glaring" low end deficiency, such that even your grandmother would agree that the converter was broken. A loss of the bottom octave or minus 6 db @ 50 hz were a couple ways it was characterized. Even allowing for the entertaining hyperbole, the claim was that there was a palpable loss of low end that pretty much anybody could hear.
-R
Post by: Extreme Mixing on November 20, 2005, 03:14:07 AM
Steve
Post by: Barry Hufker on November 20, 2005, 11:19:14 AM
I believe 7 had, for me, the best sense of space, inner detail, punch and high frequency clarity.
Barry
EDIT: I should have said also that 8 and 9 were also very good.
After a while, it all becomes a blur...
Post by: smazur on November 20, 2005, 03:43:46 PM
3,8 and 9 have a more blooming low end and seem both wider and deeper to me. If one of these clips is the straight 2", I'll go out on a limb and say it's #3.
-Steve Mazur
Edit: On second thought, forget all of the above. With repeated listening, I arrive at a different perspective each time. I would likely fail miserably at a blind test with all nine files, as did most of the original testers, and my initial, reactionary predictions are probably completely wrong. They all sound close enough to sufficiently confuse me.
Post by: The Resonater on November 20, 2005, 07:15:28 PM
I'll put the files up over speakers this week.
Post by: Curve Dominant on November 20, 2005, 08:05:32 PM
My response:
I would like that hour of my life back.
Just kidding (sort of).
[line deleted by moderator- deemed irrelevant]
One thing I definitely noticed, in all the files: Mixerman's claim turned out to be complete rubbish.
[line deleted by moderator- deemed irrelevant]
But the bottom octave was there, albeit in varying arrays of "definition" in the various files.
Not only would the cleaning lady have not heard the "bass-loss," she most likely would not have bothered to pay attention to the programme material in the first place. WTF was Albini thinking??
[paragraph deleted by moderator- deemed irrelevant]
[line deleted by moderator- deemed irrelevant]
Post by: tmheli on November 20, 2005, 08:53:52 PM
Track 2 has the most.
Track 6 has the least.
Tracks 3,4,5,7,8, and 9 are all practically identical to each other, with less bass than 1 and 2, and more bass than 6.
The difference at any given frequency between Track 2 and Track 6 is roughly 0.3dB at the most. Many folks might argue this difference should be imperceptible.
A/B'ing the first four bars of Track 2 against the first four bars of Track 6, blind, I've been able to correctly identify as "having more bass" the one that the FFT said had the most bass, which is Track 2; I'm up to 7 correct guesses out of 8 at this point.
I have an infinite loop playing, and I know which track is the first half and which track is the second in that loop. I let it play for a minute or so, without my earbuds in, with the screen off, so I have no idea where it is in the loop. Then I listen long enough to differentiate which has a more full bottom on the kick drum (I found it easiest to identify the bottom of the kick in its relationship to the attack of the kick) and say "this is the thin one" "this is the heavier one" as it switches back and forth, until I can bring up the screen and verify if I have it right. I'm at 7/8 right now, so I have a good amount of confidence in these results as "not just lucky guessing".
As far as which sounds best or worst or whatnot, I can't get into that because I have nowhere near a reference-quality system to listen on right now.
Based on Steve's assertion that he could hear a difference, and that the difference included more bass on the analog tape, I will posit that Tracks 1 and 2 represent the analog tape.
Scott
Post by: djui5 on November 20, 2005, 09:00:10 PM
| smazur wrote on Sun, 20 November 2005 13:43 |
I predict the following are the bass-light, Digi converters: 1,2,4,5,6,7. The bass guitar has a slightly nasal, pointy, "boingy", artificial quality in these clips. The guitars are not as lush and creamy. 3,8 and 9 have a more blooming low end and seem both wider and deeper to me. If one of these clips is the straight 2", I'll go out on a limb and say it's #3. -Steve Mazur |
I also though #3 to be the 2" track.
1 and 2 seemed to have a slightly, and I mean very slightly narrower and less "hyped" low end to them, and the top end is a bit extended and more open sounding.
4-7 all sounded quite close to the same to me, and 8 and 9 seem to have a slightly, very slightly wider image and more full bottom end.
The differences are so minute though, it's not wonder a lot of people didn't "hear it" or didn't speak up about it.
I have yet to listen in a studio..this is just my initial reaction from my home "reference" environment.
Another thing to consider, even though there is a slight difference in the low end, none of these files have a low end "problem" other than the slight smearing of the low end that Steve described. Possibly another reason no-one raised their hands. Just because you hear a slight difference, dosen't mean it's a "problem".
Post by: rnicklaus on November 20, 2005, 10:59:20 PM
| tmheli wrote on Sun, 20 November 2005 17:53 |
Based on Steve's assertion that he could hear a difference, and that the difference included more bass on the analog tape, I will posit that Tracks 1 and 2 represent the analog tape. Scott |
Are there 2 wave files of the analog pass? My understanding is there is only 1 track of each.
Post by: tmheli on November 20, 2005, 11:13:07 PM
| R.Nicklaus wrote on Sun, 20 November 2005 21:59 |
Are there 2 wave files of the analog pass? My understanding is there is only 1 track of each. |
I'm just hedging my bet a little.
I don't know nearly enough about any of the digital recording technologies at play here to guess which one would be the best to reproduce the low end. I guess knowing the source of Track 1 would answer that question, assuming Track 2 is indeed the analog tape.Scott
***EDIT*** 11/21/2005 01:59AM
SPOILERS. There may be stuff I'm about to say that will bias the listening process more than it may already be biased, and more than it should. If you have yet to do it, it would be a shame to read this until after you're done listening and formulating an opinion...
I couldn't stay away from the frequency analysis like I should have... I now think I was wrong with my guess about 2 being analog and 1 being the best replication of it, in the bass range. I believe I had it backwards.
It remains that I maganed to differentiate Track2 from Track6 in 8 of 10 tries... And Track2 still had the "most bass", both subjectively in my listening to the wav files, and based on the FFT.
BUT...
From looking at the frequencies above 22K, it became clear to me which ones are the low-sampling rate (48K?) Tracks. 2, 3, 8, and 9. And based on the look of 2 relative to 3,8, and 9, I'm going to say that Track2, the one with the most bass, was 48K radar, since 3,8,and 9 have a very similar character and 2 is a bit different. So then 3,8,and 9 are 48K protools.
That leaves 1,4,5,6, and 7 for the higher sample rate tracks and the analog track. I'm picking Track1 as the analog track, since it had almost the same bass as Track2, and above 22K it doesn't sit in the same space as the others. Because Track6 was markedly different from the others (it slightly *lacked* bass) I can only guess that it's the radar track, since there's only one high-rate radar track and three protools tracks. If that makes any sense!
So here's my final answer...
Track1:analog
Track2:48K radar
Track3:48K protools
Track4:96K protools
Track5:96K protools
Track6:96K radar
Track7:96K protools
Track8:48K protools
Track9:48K protools
If I'm right, then radar very slightly bloats the bass at 48K, and thins it out at 96K. This seems odd, but since it's only tenths of a dB, *whatever*. And protools very consistently thins it (again, by only about 0.1dB) regardless of sample rate or clock.
Important to note is that in the 20-60Hz range, all of this deviation is in the tenths of a dB, so all of them, in terms of a measurable frequency-type spec, perform wonderfully accurately. Even all the way up to 22K, they all did great.
The fact that I could differentiate Track2 from Track6 80% of the time is a mystery to me. If I read someone else making that claim, I would likely not believe it until I tried it myself.
Now I *really* can't wait until the results are posted, and it turns out I have everything absolutely wrong.
Post by: azuolas on November 20, 2005, 11:29:24 PM
| R.Nicklaus wrote on Sun, 20 November 2005 21:59 | ||
Are there 2 wave files of the analog pass? My understanding is there is only 1 track of each. |
there is only one file of analog pass
Post by: Fibes on November 21, 2005, 01:53:44 PM
| Quote: |
The fact that I could differentiate Track2 from Track6 80% of the time is a mystery to me. If I read someone else making that claim, I would likely not believe it until I tried it myself. |
Track 6 seems to "jump out' the most from the pack so it doesn't surprise me that it wouldn't be 100%. It's very wooly sounding and possibly my least favorite of the bunch but it still wouldn't force me into audio suicide either.
I went into this thing thinking it was a low-mid smear/push that was causing the ruckus and by looking at #6 it could be still...
Transfers are not as straightforward as one might suggest, Bob seems to be implying that it could be anywhere and whoever decides to take this further should try to put a crosshair on the Yeti if it shows up.
I still think there are much bigger fish to fry than what we've been throwing in the pan.
Post by: azuolas on November 21, 2005, 03:16:35 PM
The 3 configurations would be:
Analog, PT SYNC 48KHz, Radar 48KHz.
Post by: azuolas on November 21, 2005, 03:19:27 PM
Azuolas
Post by: Rail Jon Rogut on November 21, 2005, 03:24:15 PM
This will completely remove the console, Lavry and Nuendo from the equation.. as well as Pro Tools and RADAR. I will use a 3rd party app to do internal mixes and post them.
Thanks
Rail
Post by: azuolas on November 21, 2005, 03:29:31 PM
Just to be clear to make this a "blind" multitrack set I would label the WAV file sets as A and B. Is 48KHz PT SYNC configuration acceptable to all involved?
Azuolas
Post by: digiengineer on November 21, 2005, 03:35:07 PM
| azuolas wrote on Mon, 21 November 2005 12:29 |
Is 48KHz PT SYNC configuration acceptable to all involved? Azuolas |
Yes, since it is the common professional setup.
Post by: Rail Jon Rogut on November 21, 2005, 03:35:52 PM
| azuolas wrote on Mon, 21 November 2005 12:29 |
Just to be clear to make this a "blind" multitrack set I would label the WAV file sets as A and B. Is 48KHz PT SYNC configuration acceptable to all involved? |
Perfectly acceptable to me.
I plan to do nothing except place the files on the same tracks in the same software with the same fader levels/mix (no effects and no automation) and bounce each to disk - labelled as A and B. I will use SAWStudio since that's an unbiased 3rd party application. I will then send the files back to Azuolas if he likes to post them on his server.
Please let us know when they're available (if you can't find a DigiDelivery account to use).
BTW - you will need to strip the BWAV chunk from the sound files to make them unidentifiable. The host application (RADAR and Pro Tools) places the software name embedded in the files otherwise. if you need software to remove the 'bext' chunk let me know (and if you have a Mac or PC).
Rail
Post by: RKrizman on November 21, 2005, 03:38:20 PM
| azuolas wrote on Mon, 21 November 2005 15:19 |
I also wanted to add that I do have the original Pro Tools transfer sessions as well as RADAR system still here at CRC. I could save all of the above if there is enough interest to run your own test from some sort of DAW into your own choice of console etc... Obviously you would not have the 2" source available. I need to know this in the next few hours as the Radar is due to be shipped back to the headquarters by the end of the day today. Azuolas |
Great, Azoulas, I'd love to have that myself. Would just snail mailing a CD-R make the most sense?
-R
Post by: rnicklaus on November 21, 2005, 03:41:00 PM
I have to say, and I may be well off base, that I don't "get" this part at all.
This takes out Lavry and Nuendo, but looses the anlog and the one constant, the original analog balance.
Post by: Rail Jon Rogut on November 21, 2005, 03:44:04 PM
| R.Nicklaus wrote on Mon, 21 November 2005 12:41 |
Is this the same 14 second snippit? I have to say, and I may be well off base, that I don't "get" this part at all. This takes out Lavry and Nuendo, but looses the anlog and the one constant, the original analog balance. |
The claim is that the RADAR doesn't lose the bottom end while Pro Tools/192 do -- remove the other equations and bounce them in the box using a 3rd party application... and the resultant files should clearly show the issue. This will give a finite answer to the main question posed here -- does the RADAR record the bottom end while the Pro Tools/192 doesn't.
The question isn't does any digitization affect the bottom end -- we should all be able to agree it does... but not to an 'egregious' degree.
Rail
Post by: azuolas on November 21, 2005, 03:46:17 PM
| R.Nicklaus wrote on Mon, 21 November 2005 14:41 |
Is this the same 14 second snippit? I have to say, and I may be well off base, that I don't "get" this part at all. This takes out Lavry and Nuendo, but looses the anlog and the one constant, the original analog balance. |
I can make it either a snippet or the full song. Its all up to you guys.
You are correct it takes out Lavry and Nuendo as well as analog. However based on the comments posted it seems that most agree that the differences are more audible before you hit the print deck. To play analog source you'd have to get in touch with Steve Albini and have a 16 track ATR head available to make this possible.
Azuolas
Post by: Rail Jon Rogut on November 21, 2005, 03:47:10 PM
Rail
Post by: azuolas on November 21, 2005, 03:48:52 PM
| Rail Jon Rogut wrote on Mon, 21 November 2005 14:35 | ||
Perfectly acceptable to me. I plan to do nothing except place the files on the same tracks in the same software with the same fader levels/mix (no effects and no automation) and bounce each to disk - labelled as A and B. I will use SAWStudio since that's an unbiased 3rd party application. I will then send the files back to Azuolas if he likes to post them on his server. Please let us know when they're available (if you can't find a DigiDelivery account to use). BTW - you will need to strip the BWAV chunk from the sound files to make them unidentifiable. The host application (RADAR and Pro Tools) places the software name embedded in the files otherwise. if you need software to remove the 'bext' chunk let me know (and if you have a Mac or PC). Rail |
BWAV chunk and creator sets will be stripped. If there's no digidelivery available for this FTP transfer itself will strip most of the BWAV header information.
Azuolas
Post by: Rail Jon Rogut on November 21, 2005, 03:51:48 PM
| azuolas wrote on Mon, 21 November 2005 12:48 |
BWAV chunk and creator sets will be stripped. If there's no digidelivery available for this FTP transfer itself will strip most of the BWAV header information. |
Nope -- an FTP transfer would only remove that data from an sd2 file. You'd actually need to remove all extra chunks from the sound files to make them unidentifiable (including the Pro Tools 'regn', 'ovwf', 'elm1', etc chunks). If you have an XP system I can write an app to remove all chunks except the 'fmt ' and 'data' chunks in an hour.
Rail
Post by: rnicklaus on November 21, 2005, 03:51:53 PM
These are NOT, by the way, public domain recordings.
Post by: Rail Jon Rogut on November 21, 2005, 03:57:36 PM
| R.Nicklaus wrote on Mon, 21 November 2005 12:51 |
I take it you have cleared this with Mr Albini? These are NOT, by the way, public domain recordings. |
That's why I was suggesting I get the source file and sending Azoulas the mixes back which he can host. It shoud be the same as he's doing now with the existing files.
If he's just sending me the files I can swear that I won't look at the files' metadata and create the test mixes and send them immediately back to him to edit and post. I would prefer not to have the metadata though to remove all doubt.
Rail
Post by: RKrizman on November 21, 2005, 03:58:04 PM
Sheesh.
I can't really go down the line and say which is which. I've never used Radar or an I/O synch, so I don't know what to listen for.
-I think #1 is the analog mix. It feels wider and more natural, with plenty of lows.
-#8, to my ears, has less bass than #1 and an overall "tickier" kick. Given the premise of the test I'd say this is PT. Less bass, I think, but in other ways more defined.
-#7 also exhibited some of that, or I'm dreaming (50/50 chance). There's some flat quality about 7 I don't like.
-#9 seems to have more low end than #8, but shaped differently from #1. Less continuous or integrated somehow. Radar?
Of all of them, #3 grooves the hardest, feels the best, for me. That's how I would have mixed it. I thought perhaps that represented the monitor path that Steve actually mixed through, but I doubt it if it's not the analog.
In short, I'm going to contrast #1 with #8, in terms of low end.
Feel free to saw on either end of this limb.
(is this the right thread for this?)
-R
Post by: dragonfly on November 22, 2005, 02:53:32 PM
2. Protools
3. RADAR
I don't know if the time line position of these 3 files is the same as the 9 files samples but if it is that would mean
1 in the THREE test = 1 in the NINE test
2 in the THREE test = 3 in the NINE test
3 in the THREE test = 8 in the NINE test
but maybe they are switched in time positions between the two tests to make it more difficult ? ? ? but I'm assuming that they are the same. That being said, my guess in this NINE samples test:
1. 2" tape
3. Protools
8. RADAR
I haven't followed closely what all the different clocking & sample rate tests and options were and I don't know how those would affect the samples but to me 3, 9 and maybe 5 sounded the most similar to number 2. Less bass impact & more grating highs. 7,6, & maybe 4 sounded a bit more natural with more lows but it's really hard to tell with so many files. Anyway it's been fun. Thanks all. It will sure be interesting to actually find out what is really what.