Hi for all
I am very confused, and need some help and feedback from yours the experts...
I made many listening tests; with the intention of to choose of best lossy encoder for my personal use.
I made a lot the ABX tests with Lame 3.96.1 and Vorbis 1.1RC1. I tested Lame with --preset standard setting; and for me is excelent; I can't ABX from original.
After, the turn was for Vorbis.
I tested the official version here in HA (1.1RC1). I tried first with Q2 (not transparent for me). I tried various settings; and I choose Q5 (with ITP=4), for me was transparent; and the best solution as personal encoder.
But; one day, a friend told me about of the "excelent" quality of Xing encoder... and I answered: are you crazy?? Xing is supposed not good (as reference in many internet sites);
and my friend told me: USES only, and only the Xing encoder in AudioCatalyst version 2.1; in VBR mode, highest quality; other settings are not good.
And I answered: Oookaaaayyy !!!! I will try
And I prepared (again) my source wav files (rip from original CDs) to encode with this new test encoder to my new ABX tests... and the results for me was very strange... In many files (aprox. 90% of them, total of 20 songs of different music) I could not ABX from original !!!!
I believe in the first moment: my ears was guilty !!! but... not...
I hear very easy the artifacts in for example 128 and 160 mp3s, as too, mp3s with 16 Khz lowpass cutoff. My encoder in the past (2 years ago) was FHG Audioactive in 256 CBR mode, because in 128 and 160 I listened artifacts; 192 was good; but a little artifacts from time to time appeared.
With this evidence my ears are not guilty!! Audio equipment and source wav files??? the same than Lame and Vorbis ABX tests...
Before, my opinion was: Xing is bad!!, but; now... I don't know. My ABX tests told me the contrary... Xing is supposed... GOOD? I believe that... YES.
VERY IMPORTANT: I tested only the Xing encoder from AudioCatalyst v2.1; VBR mode, highest quality... I don't have any tests of other versions of this codec..
The test was only for music files; I didn't test killer samples...
Please any feedback was very appreciated
Thanks in advance
This would have been fairly big news a year ago, but it was discovered that the reports of Xing being a terrible encoder were greatly exaggerated in the last MP3 @ 128Kb/s Listening Test (http://rjamorim.com/test/mp3-128/results.html). As you can see from that link (overall ratings are at the bottom) Xing came 3rd and nearly tied with AudioActive. I think you may find that Xing does not handle microattacks, such as the infamous Castanets (http://www.pcabx.com/product/reference/castanets.wav), very well, not that very many encoders do but due to the fact that Xing does not use short block switching I would expect it to handle this sample particularly poorly. However if you aren't terribly sensitive to pre-echo you may not find too much at all at fault with the Xing encoder. It certainly isn't so bad as Blade or Qdesign, which are apparently far more deserving of their mention in the proverbial hall of shame .
This would have been fairly big news a year ago, but it was discovered that the reports of Xing being a terrible encoder were greatly exaggerated in the last MP3 @ 128Kb/s Listening Test (http://rjamorim.com/test/mp3-128/results.html). As you can see from that link (overall ratings are at the bottom) Xing came 3rd and nearly tied with AudioActive. I think you may find that Xing does not handle microattacks, such as the infamous Castanets (http://www.pcabx.com/product/reference/castanets.wav), very well, not that very many encoders do but due to the fact that Xing does not use short block switching I would expect it to handle this sample particularly poorly. However if you aren't terribly sensitive to pre-echo you may not find too much at all at fault with the Xing encoder. It certainly isn't so bad as Blade or Qdesign, which are apparently far more deserving of their mention in the proverbial hall of shame .
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=264711")
Thanks for the reply
Very interesting Listening test (roberto's test).. Xing in 3rd position nearly with Audioactive... as I can see here, I am not crazy
but... I am not sure with... Pre-echo?? What is exactly this?
A few days ago in this thread:
[a href="http://www.hydrogenaudio.org/forums/index.php?showtopic=30230&hl=bloch]http://www.hydrogenaudio.org/forums/index....=30230&hl=bloch[/url]
Guruboolez upload a problematic sample for Vorbis... I download the file and after that I encoded, I listened the whitenoise that degrade the quality sound of this sample... Is this a type of pre echo? Please correct me if I am wrong...
On The other hand, tomorrow I will check with castanets , in this moment I am very tired, tomorrow I come back with the results... Thanks a lot!!
But... one detail... this Roberto test was only for 128kbps, not for high bitrates as this that I checked... I checked VBR highest quality
but... I am not sure with... Pre-echo?? What is exactly this?
[a href="index.php?act=findpost&pid=264714"][{POST_SNAPBACK}][/a]
pre-echo means that you can hear something (an echo) of a sound
BEFORE it actually occurs in the source.. Usuallly noticeable in percussive sounds like the infamous castinets.
In a lot of discussions about Xing you see distinctions between "old" and "new" Xing. I think the majority of the bad reputation came from the old version.
When I started encoding mp3s Xing was very welcome in that it was relatively fast.. at a time when "fast" still took a few hours to encode a CD (compared to a few minutes now).
I have seen a killer sample (not music) that makes Xing just give up and squeal like a stuck pig, but I've never had that happen with real music.
but... I am not sure with... Pre-echo?? What is exactly this?
[a href="index.php?act=findpost&pid=264714"][{POST_SNAPBACK}][/a]
I have seen a killer sample (not music) that makes Xing just give up and squeal like a stuck pig, but I've never had that happen with real music.
[a href="index.php?act=findpost&pid=264769"][{POST_SNAPBACK}][/a]
I have a few tunes that make LAME squeal like a horny toad too. It happens with electronic metal. LAME just gives in and cries like a baby. Can't remember which version it was and if that would occur with the latest release. It was a post-3.90.3 version for sure.
Besides, no one has really tested the latest Xing version which supposedly is better than the one i Roberto's test.
edit: typo
Hmmm.. very interesting report. I am very impressed with this.
I made a casual test, and here my results:
TEST 1): MANOWAR - Fighting the world (1987) [heavy metal sample]
Original WAV versus Xing MP3: ABX failed; transparent for me.
TEST 2): CHIELI MINUCCI - The Sun Will Always Shine (1996)
[Smooth Jazz with vocals sample]
Original WAV versus Xing MP3. I obtained: [2/8] transparent for me, too
TWO KILLER SAMPLES:
fatboy.wav: easy ABX
badvilbel.wav: easy ABX, too
But, for Lame 3.96.1; setting: --preset cbr 320:
badvilbel.wav: easy ABX, too
I didn't hear any pre-echo in the encoded music files with Xing:
For "CHIELI MINUCCI - The Sun Will Always Shine" the cymbals was bright, clear, sharp and clean.
with the:
"MANOWAR - Fighting the world" sample the sound was very clean too.
This was only a casual test; but my first impression was good.
This version of Xing with this two music samples performance very good.
Of course, more exhaustive tests are necessary for a total conclusion, no only two music samples.
NOTES: The xing encoder was the same tested by Totuma:
Xing AudioCatalyst version 2.1, VBR highest quality.
I tested with my headphones.
I believe that my ears are good. My ears have tuned and checked audio amplifiers by more than 10 years..
Besides, no one has really tested the latest Xing version which supposedly is better than the one i Roberto's test.
[a href="index.php?act=findpost&pid=264772"][{POST_SNAPBACK}][/a]
I think you mean the latest Real release, as Xing was purchased by Real. Apparently the version in Roberto's thread is the last version called Xing.
In a lot of discussions about Xing you see distinctions between "old" and "new" Xing. I think the majority of the bad reputation came from the old version.
[a href="index.php?act=findpost&pid=264769"][{POST_SNAPBACK}][/a]
Good point... My encspot recognized my Xing tests files as: Xing (new)
Besides, no one has really tested the latest Xing version which supposedly is better than the one i Roberto's test.
[a href="index.php?act=findpost&pid=264772"][{POST_SNAPBACK}][/a]
I think you mean the latest Real release, as Xing was purchased by Real. Apparently the version in Roberto's thread is the last version called Xing.
[a href="index.php?act=findpost&pid=264818"][{POST_SNAPBACK}][/a]
Yes, Real is the name of the program but the encoder inside is still called Xing, right?
I've got now a portable player, but I started to record mp3 of my CD collection more than 5 years ago.
I registered AudioCatalist 1.5 then and used to rip all my classical music, and later switched to the AudioCatalist 2.
I always had a good impression of it, and remember when 5 yars ago I was alone supporting it, getting insults from many people.
They had tested the 1.0 (you know the rule about any 1.0 software version) and had a lot of prejudice when 1.5 came out. Then you had not the Lame we know today, only a poor Blade, and Fraunhofer at a steep price.
So the choiche was one. Xing at 30$!
The funnny side is that now that I got the portable player, I started to read about mp3 again and got here, and discovered what an ABX test is: I was doing similar blind test myself then and I was supporting Xing against people that were only trusting Fraunhofer with no real ABX tests, just prejudice.
It's probably not too surprising since you are using Xing at the highest VBR setting, where the bitrates are very high. Though Xing may not use short blocks, the block length only affects the efficiency of the coding. Throw enough bits at a long block, and the transient coding may be quite acceptable, I assume. It's only towards the average VBR settings (like 50) where pre-echo may start becoming a problem, though Xing does try hard to throw bits at the complicated sections. But that's just speculation. An interesting test would be to do listening tests for each VBR setting.
Xing also encodes high frequencies (up to 20 kHz) on certain bright sections of the music and it seems to be only activated for VBR. I've always been curious about whether it is helps to leave it on or leave it off.
I still use Xing 1.5 for encoding audio for my 1 CD divx rips. I just don't care about pre-echo when watching a movie, plus I want my mp3 FAST.
edit: added 'divx'
The trouble I had with Xing files wasn't necessarily the fidelity of intelligible audio. Using an older encoder at 128kbps could have been more to blame for the bland sound (I haven't bothered trying newer LAME at 128k). The trouble was that about 1 in 10-15 songs would have annoying pops, blips, zips, or chirps. It's also possible the encoder has been made better more recently than I've had a Cyrix PR150, as well .
The trouble I had with Xing files wasn't necessarily the fidelity of intelligible audio. Using an older encoder at 128kbps could have been more to blame for the bland sound (I haven't bothered trying newer LAME at 128k). The trouble was that about 1 in 10-15 songs would have annoying pops, blips, zips, or chirps. It's also possible the encoder has been made better more recently than I've had a Cyrix PR150, as well .
[a href="index.php?act=findpost&pid=264973"][{POST_SNAPBACK}][/a]
Were you using AudioCatalyst for cd ripping or just the Xing encoder on some wave files? Those pops and blips could be a result of cd ripping errors.
As I recall, using Audio Cat on a P165 I had to do a lot of parameter twiddling to avoid ripping errors.
Hi ;
Thanks a lot for everyone for your replies, for me are very important.
I made a new tests; but now with a few killer samples, my results:
fatboy.wav: very easy ABX (same as GrayWolf);
badvilbel.wav: very easy ABX (same as GrayWolf);
velvet.wav: I can't ABX ; transparent for me.
castanets.wav: a little hard; noticeable pre-echo in this sample.
Please, any comments will be appreciated;
Thanks a lot
Settings: always the same: VBR mode, highest quality; AudioCatalyst v2.1
I'm really curious how the newest Real tuned Xing encoder perform. This encoder introduced short-block old ones lack.
I'm really curious how the newest Real tuned Xing encoder perform. This encoder introduced short-block old ones lack.
[a href="index.php?act=findpost&pid=265182"][{POST_SNAPBACK}][/a]
I have the same curiosity
Nobody has explicitly mentioned the average bitrates that this setting produces. What are we talking about here? Mid-200's, I guess...
Nobody has explicitly mentioned the average bitrates that this setting produces. What are we talking about here? Mid-200's, I guess...
[a href="index.php?act=findpost&pid=265226"][{POST_SNAPBACK}][/a]
The average bitrate of my two tests music files was:
"MANOWAR - Fighting the world" [Heavy Metal sample]: 193 kbps
"CHIELI MINUCCI - The Sun Will Always Shine" [Smooth Jazz sample]: 212 kbps
I encoded a entire album:
DREAM THEATER - Images And Words [Progressive Rock]
Track 1: 201 kbps
Track 2: 203 kbps
Track 3: 196 kbps
Track 4: 195 kbps
Track 5: 194 kbps
Track 6: 195 kbps
Track 7: 177 kbps
Track 8: 200 kbps
so that would be the equivalent for LAME --preset standard...
Nobody has explicitly mentioned the average bitrates that this setting produces. What are we talking about here? Mid-200's, I guess...
[a href="index.php?act=findpost&pid=265226"][{POST_SNAPBACK}][/a]
The average bitrate of my ABX tests Xing files was around 170 to 230
As I recall, using Audio Cat on a P165 I had to do a lot of parameter twiddling to avoid ripping errors.
[a href="index.php?act=findpost&pid=265033"][{POST_SNAPBACK}][/a]
I only use EAC in secure for rip my CDs to wav files.
In my tests I only used the encoder in AudioCatalyst, not the rip function.
As I recall, using Audio Cat on a P165 I had to do a lot of parameter twiddling to avoid ripping errors.
[a href="index.php?act=findpost&pid=265033"][{POST_SNAPBACK}][/a]
I only use EAC in secure for rip my CDs to wav files.
In my tests I only used the encoder in AudioCatalyst, not the rip function.
[a href="index.php?act=findpost&pid=265561"][{POST_SNAPBACK}][/a]
Cerbie and DonP were talking about old versions of AudioCatalyst on very old computers. AFAIK those staments don't relate to modern day realities. However I would still use EAC .
The trouble I had with Xing files wasn't necessarily the fidelity of intelligible audio. Using an older encoder at 128kbps could have been more to blame for the bland sound (I haven't bothered trying newer LAME at 128k). The trouble was that about 1 in 10-15 songs would have annoying pops, blips, zips, or chirps. It's also possible the encoder has been made better more recently than I've had a Cyrix PR150, as well .
[a href="index.php?act=findpost&pid=264973"][{POST_SNAPBACK}][/a]
I didn't hear any pops, blips, zips, or chips in my 20 Xing tests files.
The sound always was clear and free of this errors.
I believe that was cd ripping erros.
Cerbie and DonP were talking about old versions of AudioCatalyst on very old computers. AFAIK those staments don't relate to modern day realities. However I would still use EAC .
[a href="index.php?act=findpost&pid=265567"][{POST_SNAPBACK}][/a]
I know it ; but only I wanted to clarified it.
A couple of years ago ... I think it was around April 2000, where I used AudioCatalyst 2.1 in Highest Quality VBR (most people at that time thought/told me that it was stupid to use, as some players etc. could'nt play the encoded files, etc. etc., but I ignored them, and I am glad I did).
I have reviewed all the hundreds of CD's encoded and the avg. bitrate for all tracks was found to be ~178kbits/s (both rock,jazz,metal,dance,trance,new age,classical and other types exists in the collection)
I have been looking at the encoding quality reported by encspot, and most are good or "not bad", but a few are reported as "bad", often it is classical music which have been encoded at or below 100kbps (avg).
At the time of encoding I used a lot of effort cleaning each CD to avoid any possible CD reading error, and had it(audiocatalyst) set to abort if any error was encountered.
Now that Harddisc space have decreased dramatically in price since then, I was thinking of transfering all the CD's again, but only this time in a lossless format.
I have considered "MAC(monkeys audio codec)/APE, FLAC, WMAPro 9.1 and LA(Lossless Audio)
I will definately used EAC for the ripping, but I want to preserve the Image style of the CD's that EAC is capable of (using the IMG button), and saving a CUE sheet.
Now I have a few questions.
1.) Is it necessary? What is your opinion? ... is it really necessary to repeat this process (just want to hear your opinion ... I am probably going to do it anyhow)
2.) which Codec? - My favorite is LA (as it produces the smallest files), but it is less compatible (no source code) as FLAC and WMA.
3.) a.) Do there exists a Playlist format, that will allow a m3u like file to access song tracks in a CDImage.la file (much like a cue sheet)... or do I have to split it into smaller track files (this option could be the only viable ...
b.) does there exists some type of compacter, that after unpacking a .la file to .wav file will be able to reconstruct the same CDImage.wav file (exactly) as if it was made with the IMG button, that is to concatenate each of the WAV files with gaps, etc. etc. into one CDImage.wav file, and write a cue file???
thank you for your time and interest.
best regards: eskildsen