Skip to main content

Notice

Please be aware that much of the software linked to or mentioned on this forum is niche and therefore infrequently downloaded. Lots of anti-virus scanners and so-called malware detectors like to flag infrequently downloaded software as bad until it is either downloaded enough times, or its developer actually bothers with getting each individual release allow listed by every single AV vendor. You can do many people a great favor when encountering such a "problem" example by submitting them to your AV vendor for examination. For almost everything on this forum, it is a false positive.
Topic: AAC at 128kbps public listening test - OPEN (Read 19594 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

AAC at 128kbps public listening test - OPEN

Reply #25
If you are on Linux or MacOS X, be careful when preparing the samples: the scripts don't check that the samples were properly extracted before erasing the original!

So if you run the scripts while faad and flac are not in the path, you will lose the samples . Keep the zip files around just in case. 

I suppose that in DOS the batch file stops at the first error. Next time we need to make the Unix version more robust.

AAC at 128kbps public listening test - OPEN

Reply #26
Quote
Hummm..what about the anchor ?? Where is it ??

http://www.hydrogenaudio.org/forums/index....ndpost&p=183714

Quote
Anyway: since the poll here is more or less tied, simply there will be no 6th codec. IMO it's also for the better, since the test will be much less fatiguing this way.

Sorry if that decision is unpleasant to some.

AAC at 128kbps public listening test - OPEN

Reply #27
Quote
I tried sample 5, and interestingly, I get a single small crackle at the beginning (within half-a-second) of one particular sample when I set the time to 14.40 as you said.  It is consistent, meaning I can ABX based on that crackle.  It moves groups when I re-open the config file, but it's still the only file to crackle.  If I set the time to 14.00, the crackle disappears.  Weird, huh?  If I set the buffer to max, the crackle moves out a bit and becomes very faint.

Pfff... can't you try to find some more straightforward bugs? 
I can't reproduce this here (at 300ms none of the samples crackles, at 200ms all of them do). And I don't know how this could happen. I mean, it's not like there's a principal difference in playing one sample or another. Or starting playback from 14.00 vs. 14.40. God, I'm starting to hate this Java and its quirks.

AAC at 128kbps public listening test - OPEN

Reply #28
Quote
Pfff... can't you try to find some more straightforward bugs? 

The comment windows are modal. 

But sadly, I noticed the crackle bug as well: I tried Sample 6, and there was a noticeable click in all files when the range started at 13 seconds. One encoder produced a clearly different click (different position in stereo). This happened with 500 and 700 ms buffer. A buffer size lower than 400 led to dropouts / disturbed playback.

AAC at 128kbps public listening test - OPEN

Reply #29
is the old abchr somehow compatible with the java version? (config vise, - i cant install java here since i dont have admin. status for this specific machine...)

and tnx to rjamorim for setting up another interesting test, which will keep me not working on things i should 
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

AAC at 128kbps public listening test - OPEN

Reply #30
Quote
The comment windows are modal.

Hehe. That's the kind of problem I like. Not really a bug (it's somewhat intended), but I can change it in the next version, of course.

Quote
is the old abchr somehow compatible with the java version? (config vise, - i cant install java here since i dont have admin. status for this specific machine...)

No, sorry. This test uses encryption to make cheating impossible (or at least really hard). The Windows version doesn't support this, so you can't use it in this test, unfortunately.

AAC at 128kbps public listening test - OPEN

Reply #31
Quote
This test uses encryption to make cheating impossible (or at least really hard). The Windows version doesn't support this, so you can't use it in this test, unfortunately.

mkay, tnx, it is not a major problem, will just had to wait to get to another location/machine.
PANIC: CPU 1: Cache Error (unrecoverable - dcache data) Eframe = 0x90000000208cf3b8
NOTICE - cpu 0 didn't dump TLB, may be hung

AAC at 128kbps public listening test - OPEN

Reply #32
Quote
Hello.

The AAC at 128kbps v2 listening test is now open. You are invited to participate.

Instructions on how to participate are available at this page:
http://www.rjamorim.com/test/aac128v2/presentation.html

The test will run until February 28th, and may be extended if proven necessary.

Best regards;

Roberto Amorim.

I can't find any test samples.
--  Frank Klemm

AAC at 128kbps public listening test - OPEN

Reply #33
Quote
I can't find any test samples.

Links are in the readme file bundled with abc-hr_bin.zip
WavPack 4.3 -mfx5
LAME 3.97 -V5 --vbr-new --athaa-sensitivity 1

AAC at 128kbps public listening test - OPEN

Reply #34
Quote
Links are in the readme file bundled with abc-hr_bin.zip

Right. That is more or less to incentive people to actually read the readme before starting the test. Else, I'm afraid some would download the ABC/HR, download the sample packages, try to load the config files and complain nothing is working.

 

AAC at 128kbps public listening test - OPEN

Reply #35
Quote
Does the Real 10 encoder produce .m4a/mp4 files and does it encode to HE AAC?

Hi, Please see my FAQ in 'the Real 10 Platform' thread.

It answers:

Which file format is AAC wrapped in when encoded in Producer or ripped in RealPlayer 10?
The current file format for AAC is RM (.ra). RealPlayer Gold will instead rip to ADTS .aac, but the feedback concerning the preference for MP4/M4A has been forwarded, and for Gold the format will most likely change.

Why is not HE-AAC encoding supported in RealProducer, or why does not RealPlayer 10 rip music to HE-AAC at 96 kbps and below?
RealPlayer 10 does not currently rip to HE-AAC, because we did not license it for general encoding, as well as lack of HE-AAC codecs for many devices we need to support.

How can I trans-mux AAC in RM losslessly to another AAC format, for instance iPod compatible M4A?
that's easy. all you need is mkvtoolnix from
http://forum.doom9.org/showthread.php?s=&threadid=64548

1) Then run
mkvmerge -o mkafile.mka rmfile.ra

2) before
mkvextract tracks mkafile.mka 1:aacfile.aac

3) finally, if you need iPod compatible M4A, just run
mp4creator60 -aac-profile=4 -optimize -create=aacfile.aac ipodfile.m4a

You can also use graphedit with Gabest's RealMediaSplitter and a muxer. I know it works with Gabest's Matroska Muxer.

======

That said, HE-AAC encoding is supported in the Producer SDK, via a separately available plug-in, that enables Producer to encode HE-AAC. Some recent 3rd party GUIs use this feature.
Sr. Codec Engineer (video) | RealNetworks Codec Group | helixcommunity.org 
This information is provided "AS IS" with no warranties,  grants no rights, and reflects my personal opinion.

AAC at 128kbps public listening test - OPEN

Reply #36
Quote
If you are on Linux or MacOS X, be careful when preparing the samples: the scripts don't check that the samples were properly extracted before erasing the original!

I'm very sorry about that. I removed the "cleanup" section from the .sh scripts and updated the readme to make clear you must have faad and flac somewhere in %path% if you are using *nix.

The updated abc-hr_bin package is already uploaded to the hosts. Sorry for the inconvenience.

Regards;

Roberto.

AAC at 128kbps public listening test - OPEN

Reply #37
This is a little off topic, but it does relate to listening tests. Does anyone have a link to where the ITU has defined a score of 4 out of 5 to be transparent?

Thanks!

AAC at 128kbps public listening test - OPEN

Reply #38
Quote
This is a little off topic, but it does relate to listening tests. Does anyone have a link to where the ITU has defined a score of 4 out of 5 to be transparent?

I'm not sure, but if I remember correctly it's part of R BS.1116-1

http://www.itu.int/rec/recommendation.asp?...1116-1-199710-I

AAC at 128kbps public listening test - OPEN

Reply #39
Quote
This is a little off topic, but it does relate to listening tests. Does anyone have a link to where the ITU has defined a score of 4 out of 5 to be transparent?

Thanks!

The term used by the EBU is "indistinguishable"

http://faac.sourceforge.net/wiki/index.php...distinguishable

The listening tests we perform here won't be able to exactly meet this very strict standard, for one because we typically don't include the reference as another sample to be rated, and for another, we don't have the luxury of getting 40 participants.

As a general rule of thumb, though, I'd say that anything averaging > 4.0 is at least very good.

ff123

AAC at 128kbps public listening test - OPEN

Reply #40
Quote
we don't have the luxury of getting 40 participants.

Not only that luxury, but those listening test recommendations specify room dimensions, equipment, listener training, pre-screening and post-screening of listeners, etc, etc.

That's why formal listening tests conduced by the MPEG, ITU and EBU cost 5-figure dollars/euros. I try to adhere as well as possible considering my non-existant budget.

AAC at 128kbps public listening test - OPEN

Reply #41
Quote
That's why formal listening tests conduced by the MPEG, ITU and EBU cost 5-figure dollars/euros. I try to adhere as well as possible considering my non-existant budget.

hehe. and you're doing a great job, AFAICT.

[span style='font-size:8pt;line-height:100%']edit: typo (dooh!)[/span]
Nothing but a Heartache - Since I found my Baby ;)

AAC at 128kbps public listening test - OPEN

Reply #42
Quote
I try to adhere as well as possible considering my non-existant budget.

And you do very well Roberto.  I don't think we could possibly give you enough thanks for these tests.  I wish there were some way that we could get everyone to act civilized while discussing these tests and not quibble over the bitrates/settings used or the results.

edit:  Sorry for the off-topicness
gentoo ~amd64 + layman | ncmpcpp/mpd | wavpack + vorbis + lame

AAC at 128kbps public listening test - OPEN

Reply #43
I finished and mailed off my results for all 12 samples tonight. I won't say anything to anyone else considering participating in the test other than I wish you well:)
you will make mp3's for compatibility reasons.

AAC at 128kbps public listening test - OPEN

Reply #44
Hello,

I have some general (perhaps stupid) questions about running the test.

1)  Do you want us to submit ABX results (and if so, for all samples?)

2)  I am noticing the "popping" in various waves.  This would almost certainly throw off my ABX results.  Should I not ABX in these cases?

3) If we cannot perceive any difference between the reference and the two unknowns, is the appropriate procedure to simply leave them at 5.0?

All that I can think of for now...
I just discovered Opus. Holy mackerel!

AAC at 128kbps public listening test - OPEN

Reply #45
Quote
1)  Do you want us to submit ABX results (and if so, for all samples?)


Not necessary, but often a good idea if it isn't totally obvious which is the encoded sample.

Quote
2)  I am noticing the "popping" in various waves.  This would almost certainly throw off my ABX results.  Should I not ABX in these cases?


Are you sure this isn't just happening when switching samples? The Java ABC/HR pops when switching samples here, much more so than the old Windows version.

I haven't noticed any pop in the actual samples (but I haven't done them all yet, either).

Quote
3) If we cannot perceive any difference between the reference and the two unknowns, is the appropriate procedure to simply leave them at 5.0?


Yes.

AAC at 128kbps public listening test - OPEN

Reply #46
Quote
Hello,

I have some general (perhaps stupid) questions about running the test.

1)  Do you want us to submit ABX results (and if so, for all samples?)

2)  I am noticing the "popping" in various waves.  This would almost certainly throw off my ABX results.  Should I not ABX in these cases?

3) If we cannot perceive any difference between the reference and the two unknowns, is the appropriate procedure to simply leave them at 5.0?

All that I can think of for now...

I think I can probably answer these:

1) ABX results were not required by the instructions.  Imagine trying to ABX to 95% confidence 5 codecs multiplied by 12 samples.  That would be a Herculean task!  ABX is there to help you decide if you're just on the edge of hearing something or not.

2) You should not penalize an entry for having a spurious crackle.  If you can, adjust the start position so that the popping sound goes away.

3) Absolutely do not pull either slider down if you can't tell a difference between the two unknowns.  If you can't tell a difference in any of the codecs, don't pull down a slider so you can enter a comment about how you couldn't hear a difference

ff123

AAC at 128kbps public listening test - OPEN

Reply #47
Thanks for the responses everyone!
I just discovered Opus. Holy mackerel!

AAC at 128kbps public listening test - OPEN

Reply #48
Quote
Quote
I try to adhere as well as possible considering my non-existant budget.

And you do very well Roberto.  I don't think we could possibly give you enough thanks for these tests.  I wish there were some way that we could get everyone to act civilized while discussing these tests and not quibble over the bitrates/settings used or the results.

edit:  Sorry for the off-topicness

Start the 'Buy a new PC for Roberot fund'
< w o g o n e . c o m / l o l >

AAC at 128kbps public listening test - OPEN

Reply #49
Quote
Start the 'Buy a new PC for Roberot fund'

Haha. Instead, create a $10000 dollar fund that I can use to conduce a formal listening test

 
SimplePortal 1.0.0 RC1 © 2008-2021