Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Confused by results of my first ABX test... (Read 10713 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Confused by results of my first ABX test...

My first ABX test here, so be gentle with me! Basically what I've done is as follows...

I have two lossless FLAC files, one @ 24/96 the other @ 16/44.1. I have then used OggDropXPd v1.9.0 aoTuV Beta 5.7 to encode each file (-q 5). When played back the .ogg file encoded from the 24/96 FLAC sounds better. To test I used the ABX comparator in foobar2000. The results were as follows;



Do these results confirm that the .ogg file encoded from the 24/96 FLAC is indeed of better quality than the one encoded from the 16/44.1 FLAC and if so, how is this audible to the human ear? Having read the forums I was under the impression that anything above 16/44.1 was pointless. Now, possibly I'm going wrong somewhere, maybe with regards the encoding.

Any advice would be most appreciated.

[TOS #9 VIOLATION.  Clips are to be no longer than 30 seconds. Links removed.]

*EDIT*

Just realised a log file can be saved after ABXing in foobar2000! Doh!

Confused by results of my first ABX test...

Reply #1
If you are starting with two different files, the real question is, are those two any different, period?

If the 16 bit is a proper conversion to 16 bits from the 24 bit file, that is one thing. If the two source files were created by different processes, then there may be distinct aspects of each that have nothing to do with the bit depth.

The question of 16 bit vs 24 bit being distinguishable from each other only applies to one original source, made at 24 bit, with the 16 bit version derived from that. How it is derived is important as there are ways to screw up the process.

A probability of 90% that the files are distinguishable is indicative that they actually can be told apart, but one round at that confidence level is not science, that is, not a standard of good evidence. If indeed you can tell the files apart, "better quality" and "sounds better" are your subjective evaluations, not necessarily related to widely used measurable qualities such as noise and various types of distortion, or even resolution of actual musical signal.

Confused by results of my first ABX test...

Reply #2
If you are starting with two different files, the real question is, are those two any different, period?

If the 16 bit is a proper conversion to 16 bits from the 24 bit file, that is one thing. If the two source files were created by different processes, then there may be distinct aspects of each that have nothing to do with the bit depth.

The question of 16 bit vs 24 bit being distinguishable from each other only applies to one original source, made at 24 bit, with the 16 bit version derived from that. How it is derived is important as there are ways to screw up the process.


The vinyl rip was originally done at 24/96. The 16 bit conversion created from the 24/96 wav file.

A probability of 90% that the files are distinguishable is indicative that they actually can be told apart, but one round at that confidence level is not science, that is, not a standard of good evidence. If indeed you can tell the files apart, "better quality" and "sounds better" are your subjective evaluations, not necessarily related to widely used measurable qualities such as noise and various types of distortion, or even resolution of actual musical signal.


Sure. Yes, I should probably have said they appeared to sound different to me rather than "better quality" or "sounds better". Thanks for the response Andy. Appreciated.

Confused by results of my first ABX test...

Reply #3
The vinyl rip was originally done at 24/96. The 16 bit conversion created from the 24/96 wav file.

*How* was the conversion done?

What software did you use?
"Not sure what the question is, but the answer is probably no."

Confused by results of my first ABX test...

Reply #4
The vinyl rip was originally done at 24/96. The 16 bit conversion created from the 24/96 wav file.

*How* was the conversion done?

What software did you use?


These are the full details including conversion and software used. Not by myself I should add.

Recorded using a Linn Sondek LP12 turntable,
Origin Live power supply,
Linn Ittok LVII tonearm,
Goldring 1042 MM Cartridge,
into Graham Slee Gram Amp 2 Special Edition Phono Stage
Chord Cobra 3 Interconnect
to Edirol R-09HR @ 24bit / 96kHz wav
WaveLab6 for track splitting
Adobe Audition 3.0 for manual click removal
iZotope RX advanced 1.21 for resampling and dithering to 16bit / 44.1kHz
Traders Little Helper > fix sbe > flac

Confused by results of my first ABX test...

Reply #5
So, you used pirated music, probably created by pirated software, converted to 44.1 with who-knows-which-settings?
I am sorry to inform you that your test failed on so many levels
Error 404; signature server not available.

Confused by results of my first ABX test...

Reply #6
So, you used pirated music,


Not pirated. Shared.

These are my albums by the way, though digitized by somebody else.

probably created by pirated software, converted to 44.1 with who-knows-which-settings?
I am sorry to inform you that your test failed on so many levels


Again, piracy implies the ethical equivalent of attacking a ship, kidnap and murder. I would hope they used free software, though I presume they didn't.

I can see how not knowing what settings have been used is a problem though. I can always find out what settings were used, if that's helpful.

Confused by results of my first ABX test...

Reply #7
There is no need to resort to arguments over semantics.

If you acquired copyrighted material through unlawful channels then you will not earn much respect around these parts.

Confused by results of my first ABX test...

Reply #8
There is no need to resort to arguments over semantics.

If you acquired copyrighted material through unlawful channels then you will not earn much respect around these parts.


Where is it implied that I have "acquired copyrighted material through unlawful channels"? Just to be clear, I own the actual album and have had it digitized by a third-party.

I hope that clears up any confusion.


Confused by results of my first ABX test...

Reply #10
That's fine.

NB: the word "if" wasn't put in bold by accident!


Hehe.  Yes, I did notice that. Thanks for the response greynol. Appreciated.

*EDIT*

Apologies to the mods also. I noticed the links to the actual files were removed. I did wonder about that. I gather I should have just provided a snippet of the tracks and not the whole song. Again, apologies for the oversight.

Confused by results of my first ABX test...

Reply #11
Feel free to submit a 30 second clip of the 24-bit version in a lossless format to our uploads forum:
http://www.hydrogenaudio.org/forums/index.php?showforum=35

Confused by results of my first ABX test...

Reply #12
Quote
I have then used OggDropXPd v1.9.0 aoTuV Beta 5.7 to encode each file (-q 5)


I downloaded the files before the links were removed. And according to their metadata, they were encoded with "official" libvorbis (ver. 1.3.2 ?), not aoTuV: "Xiph.Org libVorbis I 20101101 (Schaufenugget)".


Confused by results of my first ABX test...

Reply #14
Quote
I have then used OggDropXPd v1.9.0 aoTuV Beta 5.7 to encode each file (-q 5)


I downloaded the files before the links were removed. And according to their metadata, they were encoded with "official" libvorbis (ver. 1.3.2 ?), not aoTuV: "Xiph.Org libVorbis I 20101101 (Schaufenugget)".


Thanks for the heads-up, lvqcl. I was sure I'd followed the correct link. I only found out about OggDropXPd and aoTuV from the Hydrogenaudio Wiki. I must have gone wrong somewhere...

*EDIT*

All sorted now, lvcql.  I have followed the (correct!) link and downloaded OggDropXPd v1.9.0 aoTuV Beta 6.03.

Confused by results of my first ABX test...

Reply #15
My first ABX test here, so be gentle with me! Basically what I've done is as follows...

I have two lossless FLAC files, one @ 24/96 the other @ 16/44.1. I have then used OggDropXPd v1.9.0 aoTuV Beta 5.7 to encode each file (-q 5). When played back the .ogg file encoded from the 24/96 FLAC sounds better. To test I used the ABX comparator in foobar2000. The results were as follows;



Do these results confirm that the .ogg file encoded from the 24/96 FLAC is indeed of better quality than the one encoded from the 16/44.1 FLAC and if so, how is this audible to the human ear? Having read the forums I was under the impression that anything above 16/44.1 was pointless. Now, possibly I'm going wrong somewhere, maybe with regards the encoding.

Any advice would be most appreciated.


The following two files demonstrate one way that you can obtain different-sounding results while encoding the same musical work by the same means from files at different sample rates.

24/96 version:



The 24/96 version downsampled to 16/44:



The obvious difference is the massive difference in peak levels.  This is a natural recording that I madeof a live performer using an  ultra-close  measurement mic in an ultra-quiet, ultra-dead room.

Because of the high peak levels in the 24/96 file otherwise subtle nonlinearities in the production path could lead to audible differences.


Confused by results of my first ABX test...

Reply #17
measurement mic

Aren't those prone to have high self-noise?


Hence the close-micing. 

These particular mics were DPA 4007 which have self-noise of  24 dB(A) re. 20 µPa, which is only about 6 dB worse than a typical general-purpose cardiod.

DPA 4007-48 Specs

The dynamic range of the recording is on the sl;ightly low side of typical, IOW in the 60 dB range.

There are very, very few microphones other than measurement mics that have flat response up to 50 KHz.  Small wavelengths calls for small diapragms, and that makes really low noise a misison impossible. Whiile they spec the 4006 cartridge at 1/2", that includes a lot of surrounding metal. The actual diaphragm is more like 1/4".  Technology that would get you 12 dB SPL equivaent noise with a 1" diaphragm gets you more like 24 dB equivalent SPL with a 1/4" diaphragm.

You can read more about these recordings at 24/96 recording page