IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Artifact detection?
sebasrodri
post Feb 6 2012, 20:35
Post #1





Group: Members
Posts: 1
Joined: 6-February 12
Member No.: 96967



To start with, I'm not an expert in sound. I found some way to "zoom" an audio file to help detection of artifacts.
What I used:
- Foobar (latest stable release from web)
- 5.1 headphone configuration (http://www.head-fi.org/t/447089/5-1-headphone-experience-foobar-configuration-for-all-stereo-music-files) [not my work]
+ Dolby Headphone Wrapper [Foobar component] (http://www.foobar2000.org/components/view/foo_dsp_dolbyhp) + DOLBYHPH.DLL
+ Channel Mixer [Foobar component] (http://skipyrich.com/wiki/Foobar2000:Channel_Mixer) (Kind of optional)
- Foobar ABX component (http://www.foobar2000.org/components/view/foo_abx)

For lesser work, download the latest foobar config by RPGWizard directly from http://www.head-fi.org/t/447089/5-1-headph...reo-music-files

I'll summarize the 5.1 config just in case.
1) Install the .dll to foobar (copy them to "components" folder)
2) Order the DSP as follows:

Channel Mixer


Dolby

* What I believe this does is that splits the two normal stereo channels to 5 (3 front and 2 rear) [this is what i earlier called "zoom"], then Dolby merge them back to stereo but emulating the 5.1 feeling.

Boost your artifact detection (Seeking perfection)
Rear channels receive the "worst" part of the audio and artifacts can be detected easily. I believe you WILL note the difference between 128kbps and bigger files.
I found out that I am able to detect the difference between a Flac and V0 mp3 (with latest lame encoder). I used Fall Out Boy's "I Slept With Someone In Fall Out Boy And All I Got Was This Stupid Song Written About Me" first seconds (http://grooveshark.com/s/I+Slept+With+Someone+In+Fall+Out+Boy+And+All+I+Got+Was+This+Stupid+Song+Written+About+Me/1SCN32?src=5) where you can only hear one guitar and voice. When I heard the rear channels, i heard mostly the voice and subtly the guitar. I raised the volume to listen clearly the background's guitar and at my 170kbps aac I found some nasty artifact.
Then I ripped the song to flac and started converting the file to Nero's latest AAC codec (1.5.4.0) and Lame's latest too (3.99.4) at VBR, and ABXed (?) them with the Flac to find where I'll stop detecting artifacts.

At 170 - 190 kbps I could detect artifacts at foreground (voice) in both AAC and MP3, with AAC having better quality than MP3.
Around 225+ kbps, MP3 distinguished at foreground artifacts (voice), but not in background; and AAC at background artifacts (guitar), but not in foreground. Both faded the artifact with greater bitrates.
MP3 (CBR) and AAC (VBR) showed transparency (voice and guitar) at 320 kbps.
* I did 10 ABX for each quality, I scored in them all (except the biggest bitrates) 9/10 or 10/10.
* For most tests I had my volume up to 70% (when I listen normally at 20%)

Conclusion
Artifact detecting is a relative concept, depends on your hearing, this configuration, the use of headphones, the auido/song you are testing (noisier songs hide better artifacts); you need to convert songs until you can't detect them anymore. I suggest trying with a song with few instruments.
AAC "hides" artifacts at background, whereas MP3 keeps them at foreground.

Final comments
* I tried QuickTime's AAC encoder too but sounded ugly compared to Nero's.
* Using headphones is crucial to finding artifacts
* Bitrate is an issue between Quality and Disk Space. As my test showed, 320 kbps is transparent compared with flac with my headphones at 70% and focusing on the background music of a channeled split song (you'll never hear that much). I believe 195 or 224 is enough.
* Try the 5.1 config for normal listening, sound great biggrin.gif
Go to the top of the page
+Quote Post
hlloyge
post Feb 6 2012, 21:27
Post #2





Group: Members
Posts: 701
Joined: 10-January 06
From: Zagreb
Member No.: 27018



...and of course, you are absolutely certain that this conversion doesn't add-up artifacts on it's own...? ABX test results, please?
I prefer to listen to stereo music with stereo speakers and headphones, the way they are meant to be listened, thank you, I don't really feel there is the need to boost encoding artifacts, or add up some more processing to the signal.

This post has been edited by hlloyge: Feb 6 2012, 21:28
Go to the top of the page
+Quote Post
IgorC
post Feb 6 2012, 21:54
Post #3





Group: Members
Posts: 1580
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



QUOTE (sebasrodri @ Feb 6 2012, 16:35) *
* I tried QuickTime's AAC encoder too but sounded ugly compared to Nero's.

How many samples have you tried?
I would suggest approx. 15 samples or more.

And some results for QuickTime vs Nero comparison:
http://listening-tests.hydrogenaudio.org/i...-96-a/index.htm
http://listening-tests.hydrogenaudio.org/i...-a/results.html
d.hatena.ne.jp/kamedo2/20111029

P.S. Replaygain should be also available as encoders can change the volume noticebly.

This post has been edited by IgorC: Feb 6 2012, 22:06
Go to the top of the page
+Quote Post
DVDdoug
post Feb 6 2012, 22:27
Post #4





Group: Members
Posts: 2712
Joined: 24-August 07
From: Silicon Valley
Member No.: 46454



QUOTE
I found some way to "zoom" an audio file to help detection of artifacts.
Why? What are you trying to accomplish?

We already know that lossy compression is lossy... Normally, the idea is to get the best sound quality under normal listening conditions. You might find that format 'A' appears to be worse than format 'B' in your special zoomed/altered test, but in fact 'A' may be better, or equal, in normal listening.

If you normally listen with all of this special processing, then yes, it might be helpful to find a format with fewer artifacts when using that processing.

We already know that lossy compression is lossy... Nornally, the idea is to get the best sound quality under normal listening conditions. You might find that format 'A' appears to be worse than format 'B' in your zoomed/altered test, but in fact 'A' may be better, or equal, in normal listening tests.

QUOTE
Rear channels receive the "worst" part of the audio and artifacts can be detected easily. I believe you WILL note the difference between 128kbps and bigger files.
I don't know if you'll get the same results, but an easy way to extract the "rear channels" is to use a Vocal Removal effect, which subtracts left from right. Almost all audio editors have this effect (including Audacity), or you can get the same effect by inverting one channel and mixing.

This post has been edited by DVDdoug: Feb 6 2012, 22:29
Go to the top of the page
+Quote Post
xnor
post Feb 6 2012, 22:45
Post #5





Group: Developer
Posts: 1016
Joined: 29-April 11
From: Austria
Member No.: 90198



I have to agree with DVDdoug. If you want to do lots of processing you should do the lossy compression after processing. Otherwise you're comparing apples and oranges.
Btw, it's well known that transcoding or put more generally, heavy processing, of lossy files will result in audible artifacts..

This post has been edited by xnor: Feb 6 2012, 22:49


--------------------
"we are having an educated and deep technical discussion"-amirm
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 21st December 2014 - 03:26