Discussion regarding problematic samples from this thread (http://www.hydrogenaudio.org/forums/index.php?showtopic=39314) should take place here.
Forgive me my ignorance, but when is a sample to be considered problematic? When it's not transparent at V2 or better settings? Or...
When its quality is unusually lower than the overall quality reached by other samples.
It might be something easy to ABX at -V2, but could also be a strong unusual distortion at 128kbps, as an example.
One pretty interesting sample is missing in the Lame problem samples thread.
Guruboolez offered it in this thread:
http://www.hydrogenaudio.org/forums/index....opic=39161&st=0 (http://www.hydrogenaudio.org/forums/index.php?showtopic=39161&st=0)
It is very easy to hear and messes up 3.90.3 aps 397b -V2 and 397b -V2 --vbr-new.
3.96 -V2 is not better also. But 3.96 -V2 --vbr-new does a really good job with it.
Maybe there can be found a relation between 396 vbrnew and 3.97 vbrnew behaving different.
It is a sample from guruboolez and not uploaded here afaik.
So if you can read this guruboolez may you add this to the Lame sample thread somehow? So more people can find and hear it.
Edit: lol! Forgot the name of it It is S53_wind_saxophone_a
I posted a link to the "herding_calls" sample. In my test it produced an audible artifact with 3.97b1 and 3.90.3 at -V2 and -V0. The artifact was more pronounced with --vbr-new.
Also, it produced a different artifact with LAME CBR presets below 192.
I found the sample transparent with CBR 192 kbps and higher, but not with any VBR setting I tried.
The previous discussion and test reports are in this thread: http://www.hydrogenaudio.org/forums/index....showtopic=37003 (http://www.hydrogenaudio.org/forums/index.php?showtopic=37003)
Pretty interesting. Furst time i hear it.
None of the 396/397 V2 settings or 3.90.3 aps can handle it but 396 -V2 --vbr-new is the best to me.
Pretty similar to the wind_saxophone sample.
Out of curiosity i encoded some files that have the same style like Birds.
A constant singing female voice in a recording that isn´t absolutely clear mastered.
Elend does such music. On the second album i tried i found it again with 3.97b2 -V2 --vbr-new
This sandpaper noise with birds is for sure not only an exception.
The short sample can be found here.
Deploration (http://www.hydrogenaudio.org/forums/index.php?showtopic=39335&view=findpost&p=347710)
Hello again here is another sample adding the noise i often hear with 3.97 -V2 --vbr-new
I tried some few music pieces with wind instrument inspired by the wind_saxophone sample of guruboolez.
Et voila! This piece of song is about 2 minutes long and adds it often. As meanwhile common 3.96 -V2 --vbr-new is pretty clean! Here a small clip:
Moon (http://www.hydrogenaudio.org/forums/index.php?showtopic=39335&view=findpost&p=348710)
As this is called a discussion thread i am just beginning to wonder myself a bit. Does nobody hear any problems in Wind_Saxophone, Deploration, Moon and maybe Birds? Maybe someone even did compare these samples to older lame versions? Deploration and Moon was found encoding exactly 3 average Albums with sounds i thought would make problems. Is my auditory canal bent?
Since these problems are pretty easy to repeat it is not a typical "killer sample" problem anymore to me.
As this is not happening near as much with 3.96 -V2 --vbr-new i´d like to hear someone elses mind.
Anyhow this is in strong relation to the next Hydrogenaudio recommended lame version people trust in since years!
As this is called a discussion thread i am just beginning to wonder myself a bit. Does nobody hear any problems in Wind_Saxophone, Deploration, Moon and maybe Birds?
[a href="index.php?act=findpost&pid=348712"][{POST_SNAPBACK}][/a]
I can't hear anything wrong with any of these samples. Except that the original is a bit noisy by itself... What should I be listening to? You say sandpaper... to me that means low to medium frequency noise. Is that correct?
As this is called a discussion thread i am just beginning to wonder myself a bit. Does nobody hear any problems in Wind_Saxophone, Deploration, Moon and maybe Birds?
[a href="index.php?act=findpost&pid=348712"][{POST_SNAPBACK}][/a]
I can't hear anything wrong with any of these samples. Except that the original is a bit noisy by itself... What should I be listening to? You say sandpaper... to me that means low to medium frequency noise. Is that correct?
[a href="index.php?act=findpost&pid=348880"][{POST_SNAPBACK}][/a]
Sandpaper, maybe is not the best description. It is an added noise like distortion. If you still have an analog telephone this noise is common.
On wind_saxophone it is very constant cause of the constant note the instrument plays in the end.
On the other samples it is pulsating with the music. Very constant in Moon_short. I would rate it into the mid frequency spectrum.
Maybe you have an equalizer set somehow to amplify the highs and it is more masked this way?
Maybe I have bad hearing, or I am very tired. However I cannot hear anything wrong from the samples.
Would you please give me some hints on what and when to focus on?
Thank you.
@Wombat:
I just listened to Birds, Deplorations and Moon encoded with 3.97b2 -V2 --vbr-new.
I can hear the distortions rather easily.
You seem to be a bit sad because not many people take care of these samples. I consider this thread a good idea of Gabriel's to collect unusually bad encoded samples so that Lame development can improve. For that purpose it's not necessary that many people confirm your results. I don't think anybody doubts what you hear.
And if there are people who can't hear the problems maybe it's because of equipment. As for that I have found it makes a big difference which headphones are used on a specific sample. I usually prefer my Sennheiser PX200 for abxing, but with your samples it was easier to me using Alessandro's MS-2.
After quite a lot of tests with problem samples I don't think any more I have to use 320kbps for good quality which I beleived I had to do during most of the last months.
So I tried to find out problem sample behavior with lower bitrates.
I used
- trumpet
- herding_calls
- atem-lied
which are the most problematic samples to me.
I used Lame 3.90/3.91 (actually 3.91) which according to my experience is expected to give the best results at the moment (but Gabriel is about to improve things a lot with current Lame development).
I used 'production parameters' which to me means lowpassing to 18600 Hz and to use the -h switch when using --abr x or -b x.
So I tested for x=192, 224 and 256
--alt-preset x --lowpass 18600
--abr x -h --lowpass 18600
-b x -h --lowpass 18600.
Results concerning the different usage of a certain bit rate:
--alt-preset 192 was clearly inferior to --abr 192. At 224 kbps the difference between the two abr modes was smaller, and at 256 kbps it was more or less neglegible.
I couldn't really tell -b x quality apart from --abr x, it was only in one case (herding_calls at 192 kbps) that I had the impression --abr x was a little bit harder to abx.
(As a sidenote: I couldn't resist testing 3.91 aps and ape on the samples. Results were as horrible as expected.)
Results concerning the different bit rates:
192 kbps:
Results were not transparent as was expected, but abxing was already not easy for me (with --abr 192 and -b 192).
224 kbps:
Not transparent either, but abxing was already difficult to very difficult.
256 kbps:
At least at the border to transparency to me. Exactly speaking: Within the effort I was willing to apply within this test I could not reliably abx. There were trials where I was pretty sure I was guessing right. So may be things change when spending a lot more effort. However I'm not sporting for abx mastership - to me there's no big difference between 'not abxable' and 'abxable but only with extreme pain'.
Taking it all together:
For practical listening situations where concentration is a lot lower than when abxing
3.91 --abr x yields very satisfying results for x >= 192, and quality scales well with x. For x >= 256 quality is more or less perfect in a practical sense (as long as the samples considered are really something like the worst samples).
@Wombat:
Do you also hear the sandpaper problem with -V2 ? (not vbr-new, just old vbr)
On most samples it is also there with plain V2 but always to a less degree. I don´t have the single samples on my mind and i am not at home to test it more specific.
Rarewares offers 3.98a3 now and the lame developers spent much work regarding these problem samples it seems.
I can´t abx birds, moon_short and deploration anymore, but i have to add that i am very tired.
herding_calls, wind_saxophone and the trumpet sample became MUCH better.
After all this is very promising and shows the devs know exactly what they´re doing
Edit: I am talking about v2 vbrnew
Good news!
Will try tonight.
Well, looks like this version is really a step ahead towards extremely good reliability.
Wombat has no problems any more with 'his' samples, and as he said already trumpet and herding_calls have become considerably better compared to former 3.97/3.98 versions. I can confirm this.
Having heard and encspotted this version I can also understand Gabriel saying VBR is not to blame. With this version it is true for trumpet and herding_calls: very high bitrate abr modes don't behave better than -V0, with trumpet VBR is aware that it has to go high with bitrate (I assume the same for herding_calls but cannot conclude this from the Encspot information).
Gabriel and whoever else has contributed have done a lot of effort in improving Lame's behavior, and they have done so with good success.
So everything's fine now? Not entirely.
At least not for trumpet or saxophone like sounds of a certain kind.
trumpet isn't very annnoying any more with v2 vbrnew, but it's still easily abxable (at a resulting bitrate of 248 kbps). Quality improves with v0 vbrnew (274 kbps), but abxing isn't very hard to do. v0 (vbrold) gives a little bit better quality to me than v0 vbrnew but it's no problem to abx.
herding_call's results are a bit better to me than trumpet's.
3.90.3 high bitrate --abr x behavior is definitely better - perfect to me for 256 kbps.
@Gabriel:
As you said: looks like the psymodel is the limiting issue.
I guess nspsytunes has a lot of advantages in the low to moderately high bitrate range, cause otherwise you woudn't use it. But why not use gpscho in the very high bitrate range, with VBR the 3.98a3 way or similar?
High bitrate might compensate for the one or other disadvantage gpsycho may have compared to nspsytune. Other than that gpsycho seems to be very good at pre-echo and attack problems which are major issues with the usual perceptual encoding techniques. Why give it away totally?
As you said: looks like the psymodel is the limiting issue.
I guess nspsytunes has a lot of advantages in the low to moderately high bitrate range, cause otherwise you woudn't use it. But why not use gpscho in the very high bitrate range, with VBR the 3.98a3 way or similar?
Developping two psychoacoustic models at once requires VERY intense work -- this isn't really feasible on spare time ...
Developping two psychoacoustic models at once requires VERY intense work -- this isn't really feasible on spare time ...
Sure, and it's easy to say so having not to do it.
Of course everything is left to the developers according to their time and according to whatever they think is right.
Anyway luckily they're still about to improve Lame. I'm sure my postings have not always been welcome, but they might also have contributed a tiny bit in thinking a bit different the one or other way. That's how progress works (and I can see this with my own job. There are times when I can't continue well my with my work [database development]. I'm used to talk with a certain colleague then. He's not an expert, but he's really in the game and asks me the right questions to make me see things in another light).
As for gpsycho development maybe there is not to be done extremely much as there is a good basis for that with Lame development up to 3.91. May be results are great already with porting 3.98a3's VBR behavior to the gpsycho implementation of 3.90.3.
Exactly speaking: Why not use two strictly different encoder technologies within one encoder of the kind: for -Vx up to -V2 use 3.98a3, for -V1 and -V0 use a VBR-improved version of 3.90.3. Things can merge together in case this should be considered good over time.
Exactly speaking: Why not use two strictly different encoder technologies within one encoder of the kind: for -Vx up to -V2 use 3.98a3, for -V1 and -V0 use a VBR-improved version of 3.90.3. Things can merge together in case this should be considered good over time.
Rather than merely "switching between 2 versions", it would be much better if the goods of the 2 are combined into a single piece of code. This is development. Now the "culprit" is found, developers should focus on fixing it and improving it, instead of spending their already limited resources on other merging work.
Rather than merely "switching between 2 versions", it would be much better if the goods of the 2 are combined into a single piece of code. This is development. Now the "culprit" is found, developers should focus on fixing it and improving it ...
The progress of recent years has been achieved within a certain framework, with nspsytune as a vital element.
So I guess it's not a short term option (and propably will never be an option) to generally use gpsycho.
The progress reached must be kept.
But development AFAIK was focussed on VBR in the moderate bit rate range.
-V1 and -V0 have never been attractive options, as ususally -V2 or lower was very good, and in case it wasn't -V1 or -V0 didn't change things essentially (roughly speaking).
So nothing is really given away in terms of 'progress reached' when giving -V1 and -V0 a different meaning and use gpsycho as a basis for these options.
As for gpsycho usage with current state of development there is a problem.
With the alpha version access to gpsycho is available. I did a short test with gpsycho and -V2 last night because 3.98a3's VBR is very promising. The result according this short test was very good.
But of course I will not use this combination for productive purposes. I even hesitate to continue the test. The reason is that as long as I don't know more details about gpsycho usage within 3.98 it's like playing roulette. All I know is gpsycho development has been discontinued, so the overall encoder behavior is undefined to me.
Sure the devs may know better, but I don't know anything about whether or not it can be considered reliable to use gpsycho within 3.98a3. I don't like playing roulette.
Because of other focus may be the devs are in a similar situation.
But as gpsycho development had been vital up to 3.90, it is an option for -V1 and -V0 to use exactly the 3.90 way of doing it (using exactly that source code), and in a first step to improve only the most obvious disadvantages (VBR behavior IMO) according to actual knowledge.
More improvements can emerge over time.
I downloaded Lame 3.98a3 from rarewares and I did some ABX tests with him.
I tested the infamous aps_killer_sample, which caused serious and annoying artifacts with Lame 3.97b2 and 3.98a2 with -V2 and -V2 --vbr-new. Even without ABX, this was noticeable.
For my surprise, the annoying problems with this sample seem to have disappeared with 3.98a3.
I obtained with 3.98a3[-V2] a score of 4/10; which is not significant; transparent for me.
Very interesting results in bitrate:
aps_killer_sample:
3.97b2[-V2] --> 182kbps (very annoying artifacts)
3.98a2[-V2] --> 182kbps (very annoying artifacts)
3.98a3[-V2] --> 205kbps (transparent)
3.97b2[-V2 --vbr-new] --> 181kbps (very annoying artifacts)
3.98a2[-V2 --vbr-new] --> 181kbps (very annoying artifacts)
3.98a3[-V2 --vbr-new] --> 205kbps (transparent)
NOTE: [-V2] in 3.98a3 is now --vbr-new; in other words, -V2 and -V2 --vbr-new are now the same; but this --vbr-new has been improved and now he already doesn't suffer of bitrate drop that caused problems with some samples... Excellent work!!
Harp40_1This old reference sample produces the worst LAME artifact I have ever heard at -b 128. The first two notes are totally destroyed. In Ivan's Nero AAC test I found the 48 kbps AAC encodings much better.
The problem is still clearly audible at - b 160 and -V3 --vbr-new.
LAME 3.97b2 @ -b 160foo_abx v1.2 report
foobar2000 v0.8.3
2006/02/06 09:08:26
File A: file://E:\test\LAME problem\harp40_1\harp40_1.wav
File B: file://E:\test\LAME problem\harp40_1\harp40_1.mp3
09:08:28 : Test started.
09:09:25 : 01/01 50.0%
09:10:43 : 02/02 25.0%
09:11:06 : 03/03 12.5%
09:11:14 : 04/04 6.3%
09:11:38 : 05/05 3.1%
09:11:46 : 06/06 1.6%
09:12:12 : 07/07 0.8%
09:12:31 : 08/08 0.4%
09:12:39 : 09/09 0.2%
09:12:50 : 10/10 0.1%
09:12:58 : Test finished.
----------
Total: 10/10 (0.1%)
LAME 3.97b @ -V3 --vbr-newfoo_abx v1.2 report
foobar2000 v0.8.3
2006/02/06 09:25:25
File A: file://E:\test\LAME problem\harp40_1\harp40_1.wav
File B: file://E:\test\LAME problem\harp40_1\harp40_1.mp3
09:25:27 : Test started.
09:25:35 : 01/01 50.0%
09:25:40 : 02/02 25.0%
09:25:56 : 03/03 12.5%
09:26:01 : 04/04 6.3%
09:26:05 : 05/05 3.1%
09:26:23 : 06/06 1.6%
09:26:31 : 07/07 0.8%
09:26:38 : 08/08 0.4%
09:26:54 : 09/09 0.2%
09:27:03 : 10/10 0.1%
09:27:04 : Test finished.
----------
Total: 10/10 (0.1%)
The sample is available for example here: ftp://ftp.tnt.uni-hannover.de/pub/MPEG/au...am/harp40_1.wav (http://ftp://ftp.tnt.uni-hannover.de/pub/MPEG/audio/sqam/harp40_1.wav)
This old reference sample produces the worst LAME artifact I have ever heard at -b 128. The first two notes are totally destroyed. In Ivan's Nero AAC test I found the 48 kbps AAC encodings much better.
harp40_1 is an old killer sample, and it is strange that nobody picked it up until now - good that you found it again, as I think most encoders choke on it - I hope it will be helpful to the LAME team.
I just tried
harp40_1
aps_killer_sample
angelic
@3.97b -V 2 --vbr-new
I didn't ABX them but i just couldn't hear anything annoying, different, or anything else. Guess my ears are just bad
Harp40_1
The problem is still clearly audible at - b 160 and -V3 --vbr-new.
[a href="index.php?act=findpost&pid=362261"][{POST_SNAPBACK}][/a]
The artefact you mentioned is still quite audible at -V2 --vbr-new.
I just ABXed it with -V2 --vbr-new (LAME 397b2): 10/10
Edit:
Additionially there is easily audible distortion on the second last note played (at aprox. 4s) at -V2 --vbr-new.
@ the Lame devs:
Within a non-problem sample context I tried an extreme lowpass method used by ChiGung together with 3.98a3:
Lame 3.98a3 -V6 --vbr-new --lowpass-width 4.5 --lowpass 13.5
I also tried this lowpassing on problem samples like trumpet.flac.
I found artefacts are significantly reduced by using such a lowpass.
Maybe this information can help further improve problem sample behavior (perhaps using a lowpass within internal quality control mechanisms?).
So far 3.98a3 is better than 3.97 on every one of my problem samples. The warbling and gurgling seems to be gone. Bitrate seems to be +10% at times but I haven't tested that many samples yet.
I just tested hapischord with 398a3 and tremolo on first note and distortion (or tremolo, hard to say) on second last note is still there (with -V2 --vbr-new).
Also the sample thunder I posted in the problem sample thread is still ABXable with 398a3 -V2 --vbr-new.
Hiya
After reading this I decided to do some extreme tests (v0) on some old problem samples (which can be obtained on the gpsycho page linked in the pinned post at the top of the forum). The results were somewhat surprising.
Test conditions where ABX using WinABX, test card audigy2 output to my amp (old rotel and boston cr7 speakers). I believe the last part to be important as most people here tend to ABX with headphones. version of lame is 397b2 and then 298a3
First file was blackbird.wav where it should be abxable (new word) when the synthdrums start instead I heard a definite difference in the lenght of the piano note (also sounded a bit smeared) the results for 397b2 settings are "-V0 --vbr-new"
WinABX v0.42 test report
02/19/2006 20:58:49
A file: C:\Documents and Settings\Andrew\My Documents\My Music\ABX\problem samples\ABX\blackbird.wav
B file: C:\Documents and Settings\Andrew\My Documents\My Music\ABX\problem samples\ABX\blackbird397.wav
Start position 00:03.0, end position 00:04.2
21:00:28 1/1 p=50.0%
21:01:09 2/2 p=25.0%
21:01:15 3/3 p=12.5%
21:01:23 4/4 p=6.2%
21:01:31 5/5 p=3.1%
21:01:43 6/6 p=1.6%
21:02:03 7/7 p=0.8%
21:02:18 8/8 p=0.4%
21:02:33 9/9 p=0.2%
21:02:37 10/10 p< 0.1%
21:02:43 11/11 p< 0.1%
21:02:50 12/12 p< 0.1%
21:02:54 13/13 p< 0.1%
21:02:58 14/14 p< 0.1%
21:03:02 15/15 p< 0.1%
21:03:58 test finished
As you can see from the speed this wasn't difficult (and this was with v0). With 398a3 I could not ABX at all showing there's been some remarkable improvements.
Next I tried fools.wav (the intro to Lemon Tree by Fools Garden a favourite of mine). The results for 397b2 -V0 --vbr-new
WinABX v0.42 test report
02/21/2006 21:35:47
A file: C:\Documents and Settings\Andrew\My Documents\My Music\ABX\problem samples\ABX\fools.wav
B file: C:\Documents and Settings\Andrew\My Documents\My Music\ABX\problem samples\ABX\fools397.wav
Start position 00:00.0, end position 00:30.9
21:38:12 1/1 p=50.0%
21:41:07 2/2 p=25.0%
21:43:05 3/3 p=12.5%
21:43:55 4/4 p=6.2%
21:46:16 5/5 p=3.1%
21:48:05 6/6 p=1.6%
21:48:41 7/7 p=0.8%
21:49:05 7/8 p=3.5%
21:49:33 8/9 p=2.0%
21:51:05 9/10 p=1.1%
21:51:53 10/11 p=0.6%
21:52:25 11/12 p=0.3%
21:52:34 12/13 p=0.2%
21:52:45 13/14 p< 0.1%
21:53:27 14/15 p< 0.1%
21:54:02 test finished
Took me a lot longer and gave me a headache and I was overconfidednt on test 8. The interesting thing is that again I was able to ABX for a different reason than expected (smearing an pre-echo). I noticed that right from the first two notes the stero seperation for the plucking sound was wide on the original on the first two notes and then centered for the next two. The mp3 was less wide on the first two and actually seemed to go wider on the next two. Believe it or not this is very hard to hear repeatedly as it requires a great deal of concentration (for me) but I think almost anyone with normal hearing should be able to pick this up. However it's very hard to hear with headphones and this is why it worries me that most people ABX with headphones as this is not my standard listing conditions (and I suspect others too) so keep it in mind when doing the next set of tests.
As a side note I read that there had been some sucess improving encoding by using the the --athaa-sensitivity 1 (although I think this reduced ringing). However in this case it made it even easier to pick up (I won't lenghthen the mail with the test but it took less than 5 min) so beware when using non advised settings as they can make things worse.
Once again I could not ABX on 398a3 which puts a big fat smile on my face (well done guys).
As a final note I tried to test the old favourite 390.3 with fools I got to 18/23 then 34/49 before getting too tired to go on (1h13 is too much for me) I know it's not conclusive but with time I'm sure I could show that this difference existed then too which puts 398a3 miles ahead in this case. Hope some of you can replicate results and get some use out of this
Zster
Hi,
I did manage to ABX that harp40_1 aps_killer_sample, but the artifacts were barely noticeable. I wouldn't have noticed anything unless, like in this case, I knew that this sample had a problem.
Oh, and I was with the following settings :
@3.97b -V 2 --vbr-new
I've found something extraordinary with the sample "scooter.flac" I've posted in the Samples thread. When encoded with lame with ANY setting (tested --preset standard, --preset insane, ABR 170kbps), the bass is positioned in the center while in the original it is just slightly to the left channel. Even more strange is that when encoded with latest FhG codec (found in Audioactive), it also alters the stereo image (only tested at 128 and 160k CBR).
Is it possible it has something to do with decoder used (Winamp) ???
J.M.
I've found something extraordinary with the sample "scooter.flac" I've posted in the Samples thread. When encoded with lame with ANY setting (tested --preset standard, --preset insane, ABR 170kbps), the bass is positioned in the center while in the original it is just slightly to the left channel. Even more strange is that when encoded with latest FhG codec (found in Audioactive), it also alters the stereo image (only tested at 128 and 160k CBR).
Is it possible it has something to do with decoder used (Winamp) ???
J.M.
This really sounds like a joint stereo problem.
Do you mind trying good old lame 3.90.3 with --abr 270 -h or whatever you like (but no --alt-presets)?
--abr x or -bx uses plain stereo mode and GPSYCHO as the psy model which I consider to be very robust towards problematic cases when using bitrates >= 224 kbps.
hmm.. strange, just tried 3.90.3 -b 320 -m s, and it is still here apart from the altered stereo image the bass apears to be just very slightly lower in frequency (about 1/10th of a tone), and has also slightly different sound. Also the hihats appears a bit altered and wrongly positioned. I'm starting to believe Winamp has some lo-fi mpeg decoder.
Just tried WMA CBR 192kbps... sounds exactly like the original isn't it possible that there is some flaw in the mpeg format that inhibits the encoder from encoding it properly ?
Just tried to abx scooter.flac in foobar using Lame 3.97b2 --preset standard and Lame 3.90.3 -b 320 -m s.
I cannot hear a problem with bass positioning even when directly switching from original to encoding and vice versa. But when switching I also 'felt' most of the time bass is slightly different when encoded with 3.90.3 (sounds a tiny amount fatter to me). Tried to abx but did not succeed.
As you have a suspicion that winamp decoding is faulty I tried this too (v5.2). Couldn't hear a problem but this may be due to the fact that direct comparison is impossible and my musical memory isn't good enough for such a comparison.
BTW to me bass is positioned close to the middle but slightly to the right.
I've encoded it again and to be honest I can't hear any difference anymore! This is really confusing me, I was able to clearly hear the difference before. Maybe just listening condidions changed or maybe I've encoded the mp3 from a wrong (already encoded) file. Forget it. (but just until I will be able to hear it again! )
J.M.
I've encoded it again and to be honest I can't hear any difference anymore! This is really confusing me, I was able to clearly hear the difference before. Maybe just listening condidions changed or maybe I've encoded the mp3 from a wrong (already encoded) file. Forget it. (but just until I will be able to hear it again! )
J.M.
Could well be the listining conditions or possibly your state of alertness. I noticed that stereo effects are easier to hear when playing through speakers rather than headphones (although it really shouldn't be the case). I also noticed that I was able to ABX such effects better in the evening than the morning. Odd huh? I'll try you sample out this weekend.
Hey, I've found the problem! It is the Winamp decoder as I thought. When the mp3 is played in Winamp, it sounds different, but when the mp3 is decoded to wav (using dbPowerAmp), it sounds the same as the original (except the beginning where lame adds slight pre-echo "bonus")
J.M.
Sorry for doublepost, but I think this is interesting:
The scooter sample is easily ABXable even with 3.98a3 --preset extreme
(and even with crappy headphones; focus on pre-echo on claps):
WinABX:
WinABX v0.42 test report
04/22/2006 13:55:01
A file: E:\AUDIO\scooter.wav
B file: E:\AUDIO\scooter.mp3.wav
Start position 00:00.0, end position 05:08.7
14:09:59 1/1 p=50.0%
14:10:05 2/2 p=25.0%
14:10:10 3/3 p=12.5%
14:10:15 4/4 p=6.2%
14:10:20 5/5 p=3.1%
14:10:25 6/6 p=1.6%
14:10:30 7/7 p=0.8%
14:10:35 8/8 p=0.4%
14:10:41 9/9 p=0.2%
14:10:46 10/10 p< 0.1%
Edit: Fastenc at 128k has less pre-echo than Lame3.98a3 --preset extreme
Sorry for doublepost, but I think this is interesting:
The scooter sample is easily ABXable even with 3.98a3 --preset extreme
(and even with crappy headphones; focus on pre-echo on claps):
...
Edit: Fastenc at 128k has less pre-echo than Lame3.98a3 --preset extreme
Do you mind trying 3.90.3 --abr 270 -h? (Sorry, can't do it myself as I'm not sensitive to pre-echo).
Do you mind trying 3.90.3 --abr 270 -h? (Sorry, can't do it myself as I'm not sensitive to pre-echo).
man that is nasty! the pre-echo is MUCH worse than with 3.98a3 extreme! ABX test is quite useless here, but anyways here you go:
WinABX v0.42 test report
04/22/2006 15:53:25
A file: E:\AUDIO\scooter.wav
B file: E:\AUDIO\scooter.abr270.wav
Start position 00:00.0, end position 05:08.7
15:54:21 1/1 p=50.0%
15:54:24 2/2 p=25.0%
15:54:26 3/3 p=12.5%
15:54:28 4/4 p=6.2%
15:54:30 5/5 p=3.1%
15:54:32 6/6 p=1.6%
15:54:34 7/7 p=0.8%
15:54:36 8/8 p=0.4%
15:54:38 9/9 p=0.2%
15:54:41 10/10 p< 0.1%
BTW, 3.90.3 ape sounds a whole lot better that 3.90.3 abr270 and the bitrate is lower.
man that is nasty! the pre-echo is MUCH worse than with 3.98a3 extreme!
...
BTW, 3.90.3 ape sounds a whole lot better that 3.90.3 abr270 and the bitrate is lower.
Thanks a lot. Very interesting.
Hey, I've found the problem! It is the Winamp decoder as I thought. When the mp3 is played in Winamp, it sounds different, but when the mp3 is decoded to wav (using dbPowerAmp), it sounds the same as the original (except the beginning where lame adds slight pre-echo "bonus")
J.M.
Damn.. I had assumed that their decoder worked fine. I remember that it didn't for a while some 3-4 years ago, but then it was supposed to be fine again. I guess not....
Even 3.97b2 --preset insane IS NOT TRANSPARENT on this (scooter) sample! (pre-echo!!!)
Here's a WinABX report:
WinABX v0.42 test report
04/22/2006 22:16:12
A file: E:\AUDIO\scooter.wav
B file: E:\AUDIO\scooter.insane.wav
Start position 00:00.0, end position 05:08.7
22:16:39 1/1 p=50.0%
22:16:44 2/2 p=25.0%
22:16:49 3/3 p=12.5%
22:16:58 4/4 p=6.2%
22:17:03 5/5 p=3.1%
22:17:08 6/6 p=1.6%
22:17:14 7/7 p=0.8%
22:17:19 7/8 p=3.5%
22:17:25 8/9 p=2.0%
22:17:30 9/10 p=1.1%
well, I did miss one because I wasnt enough concentrated.
J.M.
Though pre-echo problems aren't much of concern to me your sample is as I've been so long holding up 3.90.3 GPSYCHO usage here (for other reasons).
I don't want to bother you but if it matches your interests too and as you have arrived at cbr320:
do you mind also trying a selection from these candidates: 3.90.3 api, 3.98a3 -b320, and may be also current Fraunhofer codec within WMP10, Helix -B160 -X2 -SBT450 -TX0 -HF2 (this is cbr320) and (though probably useless) 3.90.3 -b320 -h, and tell about the different quality of the outcomes.
It would give some insight towards pre-echo behavior of encoders at higherst quality setting on this previously unknown pre-echo problem sample.
ok, here's the comparsion: all Lames -b 320, fastenc VBR ~200kbps
sample: scooter.flac
3.90.3: easily noticeable pre-echo
3.93.1: same as above, probably even worse
3.96.1: probably the best, usual abx result 7/10 to 8/10 (can't reliably abx)
3.97b2: little worse than above, but much better than 3.90.3 (abx in previous post)
9.98a3: same as above
Fastenc: can't abx
Fastenc sounds perfect on all my samples, I'm starting to prefer it over Lame . And it is fast as a bonus. I've yet to find a sample where fastenc does worse than Lame. The surround encoder also handles pre-echo well but it rings (192k cbr). FHG 1.0/1.2 build 63i - sound ok on this sample but not on others (watery sound, stereo collapse).
J.M.
Thank you very much for your test.
Lame 3.90.3 -b320 uses gpsycho, --alt-preset insane uses nspsytune so if you like to I'd welcome very much if you could give api a try too.
With Fraunhofer Pro 3.3.2 (build 44) available in WMP10 I guess result is pretty much like that of the surround encoder.
fastenc sounds really interesting, I just have always prefered Lame. What version of fastenc do you use? One of those available at ReallyRarewares (called fastencc there, and v1.01 and v1.02 are available)?
I've tried 3.90.3 api, and I must say the pre-echo is gone. Sounds exactly like orig. sample.
The Fastenc encoder I've used is the one included in EasyMP3 v1.1. I use VBR because CBR sounds nasty@192kbps (probably some bug) on this sample. Fastencc102.exe CBR sounds ok with -hq switch (though slow); its VBR seems to be about the same quality as easymp3 VBR; just dont forget to add also -br switch with value at least 160 to the -vbr switch, otherwise it will lowpass at ~15.5khz (eg. fastencc102.exe <input> <output> -vbr 90 -br 160 results in ~200kbps w/ lowpass around 20khz)
Edit- fastencc102 cbr without -hq (default quality) produces the same crap as easymp3 cbr
J.M.
Thank you very much.
Glad to hear such good news about Lame (I'm a Lame fan).
But looks like I shoud give up my preference for GPSYCHO.
Can you do me a final favor and try 3.90.3 --alt-preset 270? I'd like to use that for future productive purposes.
Can you do me a final favor and try 3.90.3 --alt-preset 270? I'd like to use that for future productive purposes.
with pleasure
3.90.3 --alt-preset 270 passed my pre-echo test (can't abx). Please note that this is just pre-echo test and it might as well fail on other artifacts.
//still like fastenc better 'cause it is 3x faster
J.M.
Search on scooter.flac gives nothing. Where is sample?
Thanks a lot.
You have helped me a lot with your test on your pre-echo sample.
Sure this doesn't tell about other artifacts. But for the worst problem samples (to me) the problems are negligble to me at a bitrate of 256 kbps and above. As for them I prefered gpsycho a little bit but there has never been an essential difference to me. I just had to make up mind, and decided to use gpsycho.
But your test shows me if I take it all together that the better solution is to use --alt-preset 270.
Out of curiosity I will try fastencc 1.02 and EasyMP3 (if I can get at it) on the bad problem samples I mentioned (though I don't think I will deviate from using Lame). Thank you for the detailed information.
Search on scooter.flac gives nothing. Where is sample?
here. (http://www.hydrogenaudio.org/forums/index.php?showtopic=39314) (last post)
>halb27: glad it helped
EasyMP3 is here. (http://yahoo.pcworld.com/downloads/file_description/0,fid,22730,00.asp) I think the shareware version has some limitations though.
EasyMP3 is here. (http://yahoo.pcworld.com/downloads/file_description/0,fid,22730,00.asp) I think the shareware version has some limitations though.
Thanks for the link.
Just tested my standard-bad samples (harp40_1, herding_calls, trumpet) with the Fraunhofer codecs at cbr256. At this bitrate Lame yields a not-at-all-annoying quality (though is not transparent) to me. fastencc102 -br 256000 -hq does so as well. I also tried EasyMp3 but as you said there are restrictions (especially bad for abxing: an intro is added). From just careful listening it sounds the same to me as fastenncc102 (no surprise).
As you are sensitive towards pre-echo maybe fastenc is the better choice for you (AFAIK fastenc is famed for its good pre-echo behavior). I personally stick with lame (but 3.90.3 --alt-preset 270 --lowpass 18.6 from now on).
Thanks again for your help.
ABOUT THIS SAMPLE (http://www.hydrogenaudio.org/forums/index.php?s=&showtopic=39314&view=findpost&p=396734).
(http://audiotests.free.fr/tests/2006.05/lame398a4issue.png)
ABX log = http://audiotests.free.fr/tests/2006.05/_a3_a4.txt (http://audiotests.free.fr/tests/2006.05/_a3_a4.txt)
I compared alpha 3 ("revision 3" from 2006.02.05) with alpha 4 - which sound worse (alpha 3 is also far from transparency).
setting used: -b128
EDIT: -V5 --vbr new has no issue like this. alpha 4 jump to 183 kbps whereas alpha 3_3 is only 155 kbps (quality is on first look identical).
setting used: -b128
EDIT: -V5 --vbr new has no issue like this. alpha 4 jump to 183 kbps whereas alpha 3_3 is only 155 kbps (quality is on first look identical).
yeah, that's also my experience - only CBR is affected by this artifact. But I still dont get why it happens only sometimes (and quite rarely). (e.g. my looped scooter sample - the exactly same waveform is looped several times but it artifacts only sometimes)
J.M.
edit> quick test; adding -q 5 seems to fix the artifact
edit again> -q 4 is also fine.
When encoded with default quality, there is also a very nasty dropout on the first kick, extending down to 2kHz (!)