[span style='font-size:14pt;line-height:100%']download & compare[/span]
Source:[link removed]HE-AAC 5.1 48 Khz high quality:[link removed]-> 1
[link removed][link removed]->2
[link removed]->3
WMA9 Pro 5.1 48 Khz 24 bits:[link removed][link removed][link removed]-> 1
[link removed]-> 2
[link removed]-> 3
HE-AAC encoding: aac.dll 2.5.8.1 (Nero 6.3.0.0)
aacenc32.dll 2.6.1.1 (Nero 6.3.0.0)
besweet encoding
C:\Program Files\RMFactory\besweet\besweet.exe -core( -input G:\Mes dossiers\B.A\Les deux tours\English.ac3 -output G:\Mes dossiers\B.A\Les deux tours\English.mp4 -logfilea G:\Mes dossiers\B.A\Les deux tours\English.log ) -azid( -g max -n1 -L -3db -c normal ) -ota( -hybridgain ) -bsn( -6chnew )
WMA9 Pro encoding: AC3 -> 6 WAV with besweet and same setting that AC3 -> HE-AAC encoding
WMEncoder with 6 mono wav
[span style='font-size:14pt;line-height:100%']Encoding Speed[/span]
AC3 5.1 Duration: 2 min 58
Celeron II 1470 Mhz 384 Mo
AC3 -> HE-AAC 5.1With besweet and no azid pre-pass
for HE-AAC profil (tape, radio, internet, streaming)
Encoding duration ~2 min 20
AC3 -> WMA9 Pro 5.1With WMEncodeur
for 1 pass quality profil (10%, 25%, 50% ....)
Encoding duration ~1 min 10
for 2 pass bitrate profil
Encoding duration ~2 min 25
-----------------------------
edited by moderation: Links to copyrighted material removed.
May I ask what's your goal with starting this thread? For what reason should people "download & compare"?
Do you want people to test and report (like a poll)? - If yes, for what purpose?
Do you have some question(s)?
Do you want to show something specific with these samples?
...
BTW: The AC3 sample you offer for download (I haven't downloaded the other ones) is longer than 30 seconds. It seems to be the audio track from some Lord of the Ring trailer. AFAIK this is copyrighted material and you need a permission from the copyright holder to manipulate it (e.g. demux) and to make it publicly available. I'll remove the links in your posts until you provide a license that you're allowed to do so. According to HA Terms Of Service (http://www.hydrogenaudio.org/forums/index.php?showtopic=3974#entry149481),
fair use of short music clips (under 30 seconds) for codec testing purposes is allowed.
So you may create < 30 seconds samples to replace the ones you've linked to with.
Sagittaire,
Can you please update your thread with less than 30 seconds samples? It's gotta be interesting.
Cheers,
McoreD
I think that, what Sagittaire was trying to say (although we must admit that he didn't do a very good job in expressing it right ), is that Nero HE-AAC, especially at the lowest quality profile (tape), doesn't sound that good (a lot of nasty artifacts and lack of detail IMHO), compared to WMA 9 Pro at the same (or even a bit lower) bitrate, which btw sounds decent enough and takes half as long to encode. So, I made a legal (short) version of his ac3 file (I hope you don't mind Sagittaire) and encoded it using the latest Nero (aacenc32.dll 2.6.2.0) and Windows Media Encoder 9, to try it out for yourselves.
[span style='font-size:14pt;line-height:100%']Download & Compare (http://www.hydrogenaudio.org/forums/index.php?showtopic=20840&view=findpost&p=204079)[/span]
Kind Regards;
-George.
...that Nero HE-AAC, especially at the lowest quality profile doesn't sound that good (a lot of nasty artifacts and lack of detail IMHO), compared to WMA 9 Pro at the same (or even a bit lower) bitrate
...bear in mind, this may or may not be true for this ONE sample.
one can't make general statements about quality if results are only based on one, or very few samples.
I considering doing a multichannel listening test once, comparing WMA Pro, HE AAC, Vorbis and Real Audio. Maybe using AC3 as anchor.
But the biggest issues I ran into were lack of a comparator that can playback multichannel (AFAIK, nor ff123's nor Java ABC/HR can do that) and possible lack of enough listeners owning 5.1 systems.
...that Nero HE-AAC, especially at the lowest quality profile doesn't sound that good (a lot of nasty artifacts and lack of detail IMHO), compared to WMA 9 Pro at the same (or even a bit lower) bitrate
...bear in mind, this may or may not be true for this ONE sample.
one can't make general statements about quality if results are only based on one, or very few samples.
True. Never meant to extrapolate results, I was only refering to that specific sample, although speech with background music is common place in films.
Music sample
http://sjeng.org/ftp/work/6chNightwish.rar (http://sjeng.org/ftp/work/6chNightwish.rar)
Nero HE-AAC 2.6.2.0 VBR
MS WMA Pro 9 VBR
I considering doing a multichannel listening test once, comparing WMA Pro, HE AAC, Vorbis and Real Audio. Maybe using AC3 as anchor.
But the biggest issues I ran into were lack of a comparator that can playback multichannel (AFAIK, nor ff123's nor Java ABC/HR can do that) and possible lack of enough listeners owning 5.1 systems.
great idea!
what about downmixing the 5.1 streams to stereo? (like rear and front left into left, the same for the right channels, center and lfe on both)
that way the quality would remain untouched,
also the left/right order would be correct,
everyone (not only the 5.1 people) could join the test,
abc/hr could be used and
we could use our headphones (good for our neighbours )
what about downmixing the 5.1 streams to stereo?
I can only imagine the shower of criticism on me for choosing this way
If even, because one of the main points of doing a multichannel test is to verify if codecs can keep the surround effect well. When you downmix, you lose much of that information, no matter if you just merge channels together or use advanced algorithms like DPLII.
rja: Well, I'd just call it real world usage then
Because that is in fact how many people listen to multichannel sources. If they don't have surround systems at home they will go for second best option and use their stereo speakers or headphones. Like when you conducted your other listeningtests, I'm sure there must have been some people who used speakers instead of headphones. It's how they normally listen to the music, and I think it should count.
Well, that may not have been the main obstacle, se I look forward to the day there is multichannel playback in abc/hr.
what about downmixing the 5.1 streams to stereo?
I can only imagine the shower of criticism on me for choosing this way
If even, because one of the main points of doing a multichannel test is to verify if codecs can keep the surround effect well. When you downmix, you lose much of that information, no matter if you just merge channels together or use advanced algorithms like DPLII.
yep i know its not the best solution, but better than no comparison i think
basically i doubt that there will be enough people who would like to take the test AND have a surround system
also about the "surround effect", i wonder if that is so different between the codecs, cause basically they all have the same info in the same channels i think
so to say no matter what codec was used the same info will come out of the same speaker anyways
only the quality differs (i think so at least)
Because that is in fact how many people listen to multichannel sources.
hm yeah you could say not everyone listening to a multichannel encode has a multichannel system
also about the "surround effect", i wonder if that is so different between the codecs, cause basically they all have the same info in the same channels i think
so to say no matter what codec was used the same info will come out of the same speaker anyways
only the quality differs (i think so at least)
Hmm, not really, the channel coupling may be bad. Just listen to the WMA clip The loss of spatial detail is even audible when listening to a downmixed version.
hum i dont really get it, why does the loss of spatial info cause a worse output quality when decoding in multichannel compared to when downmixing the same file to stereo?
what about downmixing the 5.1 streams to stereo?
I can only imagine the shower of criticism on me for choosing this way
If even, because one of the main points of doing a multichannel test is to verify if codecs can keep the surround effect well. When you downmix, you lose much of that information, no matter if you just merge channels together or use advanced algorithms like DPLII.
yep i know its not the best solution, but better than no comparison i think
basically i doubt that there will be enough people who would like to take the test AND have a surround system
also about the "surround effect", i wonder if that is so different between the codecs, cause basically they all have the same info in the same channels i think
so to say no matter what codec was used the same info will come out of the same speaker anyways
only the quality differs (i think so at least)
Because that is in fact how many people listen to multichannel sources.
hm yeah you could say not everyone listening to a multichannel encode has a multichannel system
You'd be testing the downsample algorithm much more then the 5.1 encoder in that case.
You'd be testing the downsample algorithm much more then the 5.1 encoder in that case.
Standard 5.1 -> 2 channel downsampling is trivial and there will be no differences between downmixers.
(excluding downmixing to Dolby Surround and Pro Logic)