Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Lame 3.93.1 released (Read 16873 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Lame 3.93.1 released

3.93.1 has been released.

By using help and feedback by some users of this forum, this maintenance release fixes the quality bug in the cbr/abr presets.

Lame 3.93.1 released

Reply #1 –
Excellent news!

Co-operation, consultation, and an improved product.

Great work by all, and quick too!

Lame 3.93.1 released

Reply #3 –
Can't find it at Mitiok or any of the mirrors, anybody got a link?

Lame 3.93.1 released

Reply #5 –
I haven't tested this build myself so I don't know if it's "OK" for the --alt-presets.  I'm sure JohnV can comment on this though.  I do think this is a step in the right direction though btw.  Good work.

Lame 3.93.1 released

Reply #6 –
I shouldn't be and not gonna be the one and only person who says 3.93.1 is ok for --alt-presets.
Juha Laaksonheimo

Lame 3.93.1 released

Reply #7 –
Quote
I shouldn't be and not gonna be the one and only person who says 3.93.1 is ok for --alt-presets.

Well I assume that you have tested this release more than anyone else.  Plus you've had more experience with the --alt-presets than anyone else besides me.

I've already stated that I'm finished in working with LAME.  Sorry if that upsets you, but I've made my reasons clear enough and I don't expect that I will change my mind again anytime soon.  I kind of assumed that you were taking the lead here with your listening tests and trying to work with the developers... if that's not really the case then just disregard my statement.

Lame 3.93.1 released

Reply #8 –
I didn't test for fixed --alt-presets, I just wanted to check out the new medium preset.  I tried a few problem samples and here are the results (compared versus --ap-s 3.92):

fatboy.wav
medium: 248 kpbs. obvious scratching artifact
aps: 258 kpbs, sounds decent, slight scratch at the beginning.


001_awe32 20sec
medium: 183 kpbs. distortion and flange on bass line intro, and scratching on later bass drum hits
aps: 173 kpbs slightly less flange, scratching not as loud.  still bad
oddly --a-p 180 w/ lame 3.92 sounds better in some ways than both above presets with this sample.

I'll do some abx'ing later maybe.

Lame 3.93.1 released

Reply #9 –
Does anyone know if they were even attempted to be fixed?
r3mix zealot.

Lame 3.93.1 released

Reply #10 –
Gabriel:

I noticed that on this page: http://www.mp3dev.org/mp3/history.html

I am listed as creating new presets for 3.94.  Could you please remove that, along with any of the code for those presets which has been added?  None of that stuff is properly tuned and shouldn't be added to the alt-presets IMO.  Since I don't plan on working on that stuff again, I don't think they should be added along with the other alt-presets since the basic idea is that the alt-presets are supposed to be the highest quality possible.  I believe it would be giving people a false impression for them to see these things because they would assume they were high quality and extensively tested, when in reality they are not.  Thanks in advance.

BTW, I also think that it might be good to add a note to the --alt-preset help, or when encoding with the medium preset, that the medium preset has not been subjected to the same level of tuning as the other alt-presets.  Again, this is for the same reasons as I just listed.  Of course, this last part is purely your call.. but I think it is the most appropriate thing to do at least until it has been more extensively tested.

Lame 3.93.1 released

Reply #11 –
@Floyd: You have a sample where preset medium has an higher bitrate than preset standard?

@Dibrom: No problem, if you want I'll remove this part of history. In any way, I'll probably work on the lower bitrate vbr presets.
About the preset medium: yes, it's less tested than your presets. But I asked several times for testing this preset, and only have few answers. Should I permanently add a warning because users that are here have no real interest in testing it?
To my mind a better solution would be to have people really widely test it. So once again for everyone:

Please TEST the medium preset.

Lame 3.93.1 released

Reply #12 –
Quote
@Dibrom: No problem, if you want I'll remove this part of history.

Thanks

Quote
In any way, I'll probably work on the lower bitrate vbr presets.


Good luck then.  I think with some of Takehiro's newer work it should be possible to get a decent sound at lower bitrates, but it will require a lot of work and testing still.

Quote
About the preset medium: yes, it's less tested than your presets. But I asked several times for testing this preset, and only have few answers. Should I permanently add a warning because users that are here have no real interest in testing it?
To my mind a better solution would be to have people really widely test it.


Well I don't really know what to tell you here.  I think the --preset medium was added with too little testing to be thought of as having the same degree of stability of the rest of the alt-presets.  I don't think that a warning needs to stay there permenantly, because I do think that more people will test it.. especially probably now, but I do think that until then there should be at least some sort of notice.

Of course, this isn't my call really anyway... just my thoughts on the matter.

Lame 3.93.1 released

Reply #13 –
Quote
About the preset medium: yes, it's less tested than your presets. But I asked several times for testing this preset, and only have few answers. Should I permanently add a warning because users that are here have no real interest in testing it?
To my mind a better solution would be to have people really widely test it. So once again for everyone:

Please TEST the medium preset.

Heh, I think the reason why people didn't test it, was that medium preset was expected from Dibrom. But now when the situation is what it is.. anyway, I'll test it when I have time.
Juha Laaksonheimo

Lame 3.93.1 released

Reply #14 –
Quote
oddly --a-p 180 w/ lame 3.92 sounds better in some ways than both above presets with this sample.


Well, if it fails on one, it can fail on another....This makes managing LAME downloads a pain. I guess I am forced to add 3.93.1, but then perhaps I should make a separate 3.90.2 (confusing ver. number) available too. (And everyone ends up DLing 3.93...)

Lame 3.93.1 released

Reply #15 –
Thanks for the link...

Lame 3.93.1 released

Reply #16 –
shh. not so loud. do not make someone angry

@dibrom:

I can't understand your problem. prove them that sound quality has decreased.

they stated that they didn't do much to the psymodel und therefore the presets should be still valid as they were in 3.90 (and they said that abr broke in 3.93.0 by accident).

I believe that they know what they do (coz they created the thing). if they say, they didn't touch the psymodel, and therefore a decrease in audio quality could not happen by design, I believe them (unless it's proven otherwise - I know that a bug can break everything, see 3.93.0, but that's generally the risk in software).

Considering my statement above I think it's unfair to claim that you cannot recommend 3.93.1 in these forums only because it's not tested enough.

audio quality COULD be better. or the SAME. can you guarantee otherwise??? why can you recommend something by not KNOWING the difference. (because it's more probable that they broke something???).

so just TEST it or say you don't KNOW. but don't recommend a favourite LAME version until then.



thanks to gabriel, alexander, takehiro and the others for wasting a new year of time, troubles and effort in creating a new version of this great piece of software, for which they don't charge anything. 


EDIT: Sorry, after reading this thread again, I noticed that dibrom didn't recommend here anything. But I hope he will, because in the internet everywhere 3.90 is referred as dibrom's recommended.

keep on,
amano

Lame 3.93.1 released

Reply #17 –
Quote
@Floyd: You have a sample where preset medium has an higher bitrate than preset standard?

well, yes and no.  on the 001_awe32 20sec.wav sample (available somewhere on HA, probably in the sample dir) I get:
3.93.1 medium:183 kpbs
3.93.1 standard: 205 kpbs
3.92 standard: 173 kpbs

Pretty strange results.  A change of 30~ kpbs in bitrate in the standard mode doesn't inspire confidence.

Lame 3.93.1 released

Reply #18 –
Check again, here awe32_20sec with --alt-preset standard is with:
3.90.2: 205.8k
3.92: 204.0k
3.93.1: 204.9k
3.94.a4: 251.7k

Lame 3.93.1 released

Reply #19 –
thats odd..  I tried again and got 173 kpbs.  hmm,  I might still be using the p4 compile, but I'm pretty sure it was the compile that produced crc identical output.  I'll try dl'ing 3.92 again.

edit: i get 204 kpbs now.  apologies for the incorrect info.  Something is seriously messed with that p4 compile, so much for crc identical output.

Lame 3.93.1 released

Reply #20 –
I also have the 3.92 compile from Yaroslav Efimov here and there are only atomic differences to his 393.1 compile.
Seems to work as 3.92 with aps.

Wombat
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Lame 3.93.1 released

Reply #21 –
Quote
@dibrom:

I can't understand your problem. prove them that sound quality has decreased.

they stated that they didn't do much to the psymodel und therefore the presets should be still valid as they were in 3.90 (and they said that abr broke in 3.93.0 by accident).

I believe that they know what they do (coz they created the thing). if they say, they didn't touch the psymodel, and therefore a decrease in audio quality could not happen by design, I believe them (unless it's proven otherwise - I know that a bug can break everything, see 3.93.0, but that's generally the risk in software).

Please go back and read the discussion about 3.93 before it was released, and after it was released.  I'm not going to bother to get more in depth than that.

Lame 3.93.1 released

Reply #22 –
Quote
Something is seriously messed with that p4 compile, so much for crc identical output.

If you missed the discussion, there was talk about how the P4 compile used some SSE routines from GOGO which do not work "properly" or "as expected" with LAME, and so the output differs quite significantly.  It's not recommended for use at all.

Lame 3.93.1 released

Reply #23 –
Quote
Quote
Something is seriously messed with that p4 compile, so much for crc identical output.

If you missed the discussion, there was talk about how the P4 compile used some SSE routines from GOGO which do not work "properly" or "as expected" with LAME, and so the output differs quite significantly.  It's not recommended for use at all.

I guess I got confused on that..  when nebob released the compile originally, he said it was crc identical.  Then he enabled some other speed ^ stuff that screwed up the crc in the next compile.  I thought I was using the original one that was safe, but possibly I tried the experimental one and overwrote it.  In any case, to other p4 users: beware those p4 compiles, do some testing on your own before using them for anything important.

Lame 3.93.1 released

Reply #24 –
Quote
Quote
Quote
Something is seriously messed with that p4 compile, so much for crc identical output.

If you missed the discussion, there was talk about how the P4 compile used some SSE routines from GOGO which do not work "properly" or "as expected" with LAME, and so the output differs quite significantly.  It's not recommended for use at all.

I guess I got confused on that..  when nebob released the compile originally, he said it was crc identical.  Then he enabled some other speed ^ stuff that screwed up the crc in the next compile.  I thought I was using the original one that was safe, but possibly I tried the experimental one and overwrote it.  In any case, to other p4 users: beware those p4 compiles, do some testing on your own before using them for anything important.

I'm not understand  I'm have P4, and compiling lame with ICL 7.0, Dmitry compiling lame with ICL 4.5, I'm tested both versions on my system, and my compile working faster than compile from Dmitry, All rest is absolutely identical, what for then to use ICL 4.5 if it works more slowly?!
My sound hardware: Creative's Audigy 2 ;)
My music encodings: Nero AAC Codec 2.6.2.0 Profiles:Transparent,Extreme