hello. i'm ol school and still use the old versions of lame. one of the reasons i do this is because of the default lowpass filter of the 3.90.3 version. it has a transition band: 19383 Hz - 19916 Hz when using the V1 setting instead of the normal 18671 Hz - 19205 Hz transitional band used by all other versions at all settings. [that iv noticed]
my question is........if you have a .dll version of a lame encoder [3.92] and not the .exe version, how can you tell what your encoder's low pass is? Is there an Mp3 analyzer out there that's easy to use that will tell this info? when comparing the file size between the two [3.92 3.90.3] at the V1 setting, there is little difference. the 3.92 version actually makes a slightly bigger file size than the older version. i would assume it still has this same "higher" transitional band as the old one at this setting, but i'm not for sure......any responses on this would be greatly appreciated. thanx

If you can't hear the difference, why do you want a higher lowpass? Just use the recommended version.

that wasn't the question, but thanx for replying................next

Your best bet is http://www.rarewares.org/mp3.html , scroll down a bit and locate the download for lame_enc.dll modified to use INI File Setup:

I'm not in a mcdonalds, so up yours. It's a valid statement I'm making.

Try EncSpot. However, it only shows the lowpass if a Lame tag is written to the file - don't know when/if this was done in the old versions when the dll is used... I have some old 3.90.3 -V1 (dll) encoded files and the lowpass is shown as 19900 in Encspot. But as others have told you, you'd be much better off with Lame 3.97beta2 anyway
Edit: I've seen you're registered to HA since 2003 and don't use the --presets but some outrageously outdated settings with an outdated Lame version. That's...pretty odd, to put it mildly

you are talking about something that has nothing to do with this thread.......are you trolling?

Jaz and STI i thank u for your recommendations on this matter. i'm fully aware that the 3.97b2 is the recommended version by this site. i have no qualms with that, but that has nothing to do with what i'm asking in this thread. thank you very much

yes i use: -b 192 -m s -V 1 -B 320 -F -q 0 3.90.3

i have my own reasons for not using the presets. i'm not the only one at this site that has had issues with the js option that is default with the presets. and yes yes i have been advised to switch to flac ape aac and all that............most car and home audio players do not play these formats. mp3 is still the standard, so i want the highest quallity mp3's i can muster and frankly i find that the setting i use above gives me much better results [for me] than any preset that has js enabled ,which is all of them.

that's not what this thread is about.
i asked a legitimate question and immediately had the 3.97b2 version crammed down my throat. if you don't know the answer to something, it's best to just move on instead of giving advice that no one asked for................once again i thank you for answering my original question..........cheeers

Errrm.. You are asking how to set the lowpass when using the dll, I've pointed you to a modified dll that lets you configure the parameters you want to user via a .ini file. Did you check it? From your reply I feel like you didn't understood me. And as I say, it's your best bet. It is this, or stood with what you have.

sorry for the delay......i was trying to find it with the EncSpot software. apparently the 3.92 version does not put the lame tag on these mp3's because i can't find the lowpass with it.

i went to rarewares and found lame_enc.ini but it comes with the 3.97b2mod zip. i'm not really for sure what to do with this. any suggestions JAZ?

He didn't ask about changing the lowpass but about seeing what lowpass is used, that's why I recommended Encspot.
BTW, chronicking, you can disable JS also in all presets (3.90.3 - 3.97b) by adding -m s to the commandline; also the lowpass can be changed to any value.

well i know much time is put in these presets tweaking them and such. i figured if i started changing anything about {js for instance} it might throw the whole configuration of the preset off causing more probs than good......btw encspot wouldn't show me what the low pass for the 3.92 version is @ V1

Why don't you try using a 3.92 .exe, which should show the lowpass used. The .dll will use the same lowpass. Otherwise, you could try looking at a spectrogram view (in a .WAV editor) of the encoded files and approximately guess where the lowpass is set.

i wish i had a 3.92 exe. all i have is the .dll that came with cdex. thanx again i shall try to find the appropriate .wav editor.

Indeed... It looks like i can't read sometimes .
And even more, my advice does not allow changing the lowpass (I never used it, and i thought it allowed passing parameters, not only to select a preset).

Here you go: direct download link from reallyrarewares: Link

THANX MAN!!!!!! that was exactly what i needed. it created the exact same file size as the .dll that came with cdex when i used it with razorlame. and yes it uses the same lowpass filter as the 3.90.3 version @ V 1 MUCHO Glacias!

too bad the new presets don't use the same lowpass @ V1 without adjustment.....

no not bad.

Better

data for unhaerd audio was moved to describe what people actually could hear.

Like saying to bad there was no more sugar in the cake that i didn't bake. (bits used in audio not heard)
that's because he sugar was used for the cake you DID bake and eat and could enjoy (bits used to improve thing you can hear)

maybe the diabetics here [those that can't] don't like all the sugar in their cake. i on the otherhand, like my cake sweeeeeeeet, dripping with sugar baby!!


apparently someone back in 2001 liked their sugar as well.......both V1 and V0 use the 19.5khz lowpass. i also noticed that their are some sweet lovers today as well.......the 3.97b2 also use the 19.5khz lowpass as default when using -V 0.

VIVA LA SUGAR!!!!!!!!

I invite you to read the TOS: http://www.hydrogenaudio.org/forums/index.php?showtopic=3974



And please be more gentle with people that are trying to help you.

Thank you Oge_user.

I think the moral of this thread is "to each his own." While everyone is welcome to have their own personal preferences with regard to the lowpass filter, I'd like to suggest a little fun experiment - try out the tone generator feature in foobar:

In <add location>, type "tone://{freq},{dur}" w/o quotes (for example, "tone://16000,5" generates a 16000 Hz tone for 5 seconds).

I started testing at 14000 Hz and worked my way up in 500 Hz increments. At around 17000 Hz I was unable to perceive the tone. Thus, for ME, allocating bits for audio above this frequency is somewhat of a waste. The point sven_Bent was making is that these bits could be more effectively used to represent sound IN my hearing range than outside of it.

This test was one of the reasons I settled on --preset medium (which has a lowpass of 18000 IIRC) as my setting of choice.

Please stop making such condescending responses to chronicking. It should be obvious that he has read TOS#8, from the care he has taken to avoid breaking it. If you disagree, please quote anything chronicking has said, that "puts forth a statement concerning subjective sound quality", without carefully qualifying it by adding "[for me]", as in post #9.
I have reread this thread several times, and cannot find a single example of chronicking being anything but gentle. If I have missed such an example, please quote it.
Again, there is no need for such condescending statements. It is evident from his response that chronicking understood sven_Bent's point.

Regards,
Madrigal

I'm not making any condescending response, instead I only invited chronicking to read the Terms Of Service. I quoted TOS #8 because his posts are on the edge of such TOS: he put forth a statement concerning subjective sound quality without provide objective support (ABX) for his claims.


It's not good to reply like this to an user who was trying to help (follows the quote):

For example, I want to ask chronicking what kind of issues he had with Joint Stereo used by presets and if he can provide this through ABX'ing.

Please quote anything chronicking said, that "put forth a statement concerning subjective sound quality", without carefully qualifying it by adding "[for me]", as in post #9.
I disagree. Shade[ST] was not trying to help with the question at hand, but was instead trying to change the subject. For this reason his response deserved to be dismissed, and I believe chronicking accomplished that gently enough for the average adult member of a forum such as HA.

Regards,
Madrigal

Stating that he has his "own reasons for not using the presets" and has had "issues with the js option" does not constitute a violation of TOS#8. Far from putting forth any specific "statement concerning subjective sound quality", chronicking has simply expressed a personal preference for his own particular command line, however misguided that might be.

Regards,
Madrigal

In this case there's a very fine line between a statement, or a subjective opinion, or a claim (statement, subjective opinion and claim are the terms used in TOS) and a personal preference.

I don't have any intention to continue with this debate, so please stop here.

Done.

Regards,
Madrigal

i don't consider madrigal taking sides in this matter even though he is on the right side. i've talked to countless people about this site and how it is down right rude when you go against it's grain thus the recommended version of lame. i'm sure there are people here that still use wimamp ver 2.81 or the 6.4.x.x version of the classic media player. this is no different than using an older [well proven] version of lame if you are happy with it's results. i could understand the trolling or flaming against me if in someway i was recommending another version of lame besides the one that is recommended by the powers to be at this site, but i'm not and wasn't. i thought this was an informational site about audio compression. maybe you should change the name of this site 2 "LAME 3.97b2 or DIE"

i think if this thread was re-read and a look is taken back at most of the threads similiar to this one or any thread that speaks of using anything but what is recommended by this site and one would see the redundancy of this site's members continuously flaming, trolling, and bashing people that have legitimate interests and ideas other than the ones dictated at this site. this is a resourceful site but just re-read the thread and many like it. why do people that don't know continue to blurt out "just use the recommended version" ? this makes you sound like you actually know less than if you would just say "i don't know"

Have you followed the discussion surrounding the advancement of Lame 3.97b1 to becoming the recommended version? It illustrates how anal this community is about accepting changes and maintaining a high quality standard. As such it is quite an achievement this happened at all.

It is only natural, if you come to the place where this happened, that your decision is questioned. The community would be doing you a disservice would it not try to persuade you. Especially if your decisions go against what is considered common knowledge here. For example you using -q0 with 3.90.3 shows me that your choice is based on limited knowledge.

But since you now have the information you want, let's move on...

NEXT!

you can use analfreq 1.8 http://www.softlookup.com/download.asp?ID=2908&DID=4J58YURT
to see where the lowpass is.. its free, it works

boy that sure was informative!!!!! and very necessary for this thread...thanx Gecko and yes i gots it!!!

thanx man i'll give that a shot. nero's wav editor is lacking[for me ] to say the least.

Being a high bitrate 3.90.3 non-alt-preset abr lover (and a 3.97b/3.98a adorer for the low to medium bitrate range) I do know from my own posts that there are always people holding up common wisdom who feel obliged to attack posters with a different attitude.

Meanwhile I consider this being natural and if things go best it can have two positive outcomes:
- the poster who deviates from common wisdom has a chance to improve his knowledge concerning the one or other detail. Often there is no universal wisdom to those details, but it's helpful to learn about other arguments. An example for this is the lowpass discussion in this thread.
- those who hold up common wisdom have a chance to see that common wisdom usually applies within a certain framework and that this framework is not necessarily universally extendable. This makes discussion like in this thread meaningful and refreshing.

@ halb27:

Good post. Personally I would have said "challenge" rather than "attack", but in general it was very well put.

Regards,
Madrigal

so correct sir. i do respect the findings of this site and respect the developement of lame even more. even if it might be the developement done 3-4 years ago. but back to that learning thing you mentioned......if the -q0 switch is not recommended for 3.90.3 why have things such as this in the changelog::::# Takehiro Tominaga: new noise shaping mode, offering more "cutting edge" shaping according to masking, enabled via -q0 if i'm not mistaken it was reworked again for the 3.92 version [4 cdex possibly?] 0 = slowest algorithms, but potentially highest quality.........is this additional length in time not giving the encoder more time to distribute bits accurately? i've done many file SIZE tests and in some instances the -q 0 will produce a slightly bigger file and in some instances the default -q 2 will make bigger file sizes. in some cases -q 1 would be the biggest. granted -q 0 may mess up all that fine tuning done in the presets but is it not better to give the encoder more time to accurately encode the file to the best quality? especially if time is not an issue.

Personally I do not use -q0, but -h. All I've read about -q0 doesn't bring me confidence in it's usage. And I don't think like slower=better (which sometimes is, but not as a rule).
Edited later:
If you think it's worthwhile to squeeze the max out of the -q switches I think it's more reliable to use -q1.

Well, it was your choice to ignore information presented to you and even complain about it. But since you got it...

See http://www.hydrogenaudio.org/forums/index....showtopic=35916 on the q0 bug. Apparently it can cause severe ringing on some samples (examples provided in the thread).

I believe one of the things the q switch influences is the redundancy compression part. Like when you set Zip to "Best" instead of "Normal". But other settings are also modified.

An interesting thread to test your real-world cutoff point: http://www.hydrogenaudio.org/forums/index.php?showtopic=4894

The response would have a lot to do with the way information is being presented to chronicking - by ankle nipping usage minders

Now you are the one making unqualified statements about lame quality. There is no data or proper reference in that thread to q0 s performance in version 3.90.3, only 3.96 and some unqualified speculation about previous versions. Even if the relevant 'q' functions were unchanged between 3.90.3 and 3.96, 3.90.3's performance could have been unaffected, and there is yet no report of it being affected despite much testing and usage.

Finding a problem in 3.96s q0 performance, would suggest 3.90.3's performance may suffer similarly. The problem not having been picked up during 3.90.3 testing and subsequent usage would suggest the contrary.

If you wish to claim q0 does reduce quality in 3.90.3 document it please.

How do you suggest such a situation be handled in future? Should we not question someone's motives to ignore the recommendation? Both the current one and the recommended settings for version 3.90.3?

There are numerous people who come here with the misconception that they need a higher lowpass. That joint-stereo messes up the sound etc. This has been refuted on numerous occasions. Should we assume that each and everyone has done his own double blind testing to determine his individual needs which run contrary to the majority of people?

I am not saying that 3.97b2 -V2 --vbr-new is the setting for each individual. But it is for a lot and probably overkill for many. Claiming to be old school does not automatically put you in the minority which actually needs higher or "different" quality. If the first reply in this thread would have been: "Have you run double blind tests confirming that you need specifically what you are using and asking for? Or are you just acting on the warm fuzzy feeling?" I bet the reaction would have been the same even if you'd scratch the second part.


That is precisely the reason why I put the word "apparently" in there. I am only citing others. But you are right, that particular thread does not have anything which would be considered proof around here. Unfortunately I wasn't able to find a better thread, as searching for -q0 and its variants is difficult. The fact that the samples guruboolez posted are also not working is even more unfortunate. I apologize. However it can be assumed that guruboolez isn't making any vague claims without backup.


Hm, I think it's debatable whether many people* use q0. One can speculate that q0 wasn't taken into consideration very long, during testing of 3.90.x, because of its negative effects on speed and perhaps no noticable increase in quality. Thus it was not as thoroughly tested and the bug slipped through?

Anyway, I will contact guruboolez and ask for the samples to run some tests.

*There are probably many that use their homebrew q0 commandline, but I haven't seen them come here and be able to proove its superiority.

I agree that your points are relevant, me I prefer to listen to music that is 'psycho acousticly enhanced' by a wide lowpass starting at ~13 Khz -takes the edge off harsh vibes without affecting clarity or tonality to me.

'Just when someone likes to chip in a couple of sentences and dismiss a persons enquiry altogether, only tit for tat can result. Idealy a reply would provide the requested information, or at least wait until its supplied, and then explain why the information requested is troublesome. Short dismissals are more likely to entrench the offending parties preference.

I dont see why people should need to, unless explicity making a contentious claim.
The prosecution of implicit misunderstanding is the work of the 'thought police'
Ive read good threads come of the j-stereo contention, and have seen it demonstrated as the superior option for standard playback situations, but am unsure of its superiority for all requirements. Nervousness about common interpration of debates, seems to hinder enquiry there.


To me those questions would seem uncalled for until a point of contention is actually made. I dont think it was chronikings purpose here to divisively imply that higher lowpasses are generaly merited. Anyway such subtle divisiveness would be better ignored, tacticaly speaking

So chronicking could have ended his post with "apparently - the lowpass is too low and the higher, the better" and that would have been ok? In fact he avoided as well as one could suggesting anything close. Only later had his own personal lowpassing foible drawn out with the banter. He could have said he was encoding a natural soundtrack for his dog to listen to, and then no one would have a leg to stand on

No biggie, though should be easy to test with exact same samples if one has the hearing. Could be another reason to demote the trusty old 3903, especialy with latest improvements in 397&8 across all the bitrates.

Well, I tried with the 2 samples provided by guruboolez but couldn't hear any difference between the q levels. I just used a basic command line: lame -q0 sample.wav. So 3.90.3 doesn't benefit from any preset tunings.

Versions: 3.90.3 (q0 q2), 3.96.1 (q0 q1 q3), 3.97b2 (q0 q1 q3)

Then I tried one sample of my own: angelic.wav
Very much to my surprise, 3.90.3 introduced an ugly high pitched blip at q2 which was noticably reduced with q0. (ABX 8/8)

While otherwise 3.90.3 sounded OK, 3.96.1 brought in some ugly tappity noises (like when someone is crumpling the styrofoam on a microphone) which were (almost) absent with q0 but increased with each q level (also tested q2 on this one).

3.97b2 brings out prominent crumpling at q0 allready.

So much for that. I have no own evidence to support my words about q0 and I do not plan to run further tests on this subject any time soon.

@ChiGung: You make very good points. About the "apparently" issue: what I meant to say was: I have not verified the claims made in the thread I linked to myself, so I can not vouch for them. Perhaps my use of "apparently" was wrong? After all, English is not my native tongue.

The q0 bug is something which from my perception stayed below the radar of many forum regulars. From my memory (and the quoted thread supports that) there actually was a bug which the developers acknowledged and fixed during the alpha stage of 3.97. q0 in 3.97 represents something which was not even present in version 3.90.3 (but it was in 3.96.1). q0 from 3.90.3 is now q1 in 3.97b2. So I guess no claims made for 3.97 q0 can be translated to 3.90.3's q0 performance.

I think you used it correctly, normaly it can be casualy thrown in at the beginning, middle or end of a sentence to indicate a thing is apparent but not secure.
No matter, thanks for the testing and intresting report.


Thanks for the headsup.
regards'

----------------------------------
Apparently, Tom Cruise is...
Tom Cruise apparently is...
Tom Cruise is apparently...

IIRC this was well documented back when 3.90.2 came out. However, that was years ago, and its nearly impossible to search for 2 character words.

The best I could find is this oldish thread which claims to have a sample where 2 is better then 0:

http://www.hydrogenaudio.org/forums/index....showtopic=11925

And this thread where Dibrom says that -q 0 doesn't help and therefore he recommends that it not be used:

http://www.hydrogenaudio.org/forums/index....c=593&hl=switch

That last one is over four years old, and probably still being stated after the fact, so if you're interested, theres still probably more threads to be dug up.

Hallo ChiGung,

It was you who inspired me most to use high bitrate abr (instead of cbr320) and to lowpass at 18.6 kHz which I actually do with my productive encodings. Thank you very much again.

Because of this background I'd be very pleased to learn more details about your wide lowpassing. Sounds interesting.

Sorry for going a bit off-topic.

eh, you know Im into speculation, you went on and checked that idea yourself Im glad its working for you

This one is very personal. I imagine a world where lowpassing has not been possible, and people can only listen to transparent hifi. Then someone brings out a clean lowpass filter, its marketable as a technique which can electronicaly enhance sound to make it safer to listen to and less stressfull to the ears at high volume, yet not affect clarity. Because weve come from the other direction and technology has had difficulty extending the frequency range, full range is regarded as desirable, but if lowpassing was the hard thing to do, i think it would be the latest rage in hifi.

Considering, very few hearing aids (i dont need btw; ) go beyond 7Khz to assist clarity, cymbols, twangs, percussion doesnt suffer in loudness to me, until lowpassed under ~10 Khz, a sharp lowpass beyond 13Khz - I cant personaly even detect with my hearing, I use --lowpass-width 4.5 --lowpass 13.5 on my portable to ameliorate high frequency content past 9 khz.

It doesnt sound transparent to me, its a superior audio reproduction for my puposes.
With my own hearing I can happily go lower than this, but Im mindful of situations where other people listen to my playlist too.

If organising ambient cafe or party music or something, especially playing lossy encodings, I think it would be wise from a public consideration angle to use lowpassing or strongly reduce very high frequencies. Who would mind? people blessed with sensitive ears? I expect only equipment coinessures would complain about the bandwidth of the PA system, every one elses ears and atmosphere should bennefit, from 'pro-fidelity lowpass enhancement'

I cant vouch for the quality of lames' lowpass, though I havent come across any damning reports other than the apparent consensus that any detectability of lowpassing is comprimised and undesirable. Why should that be? do we really need the real thing, or even better than the real thing ?

On the other side of the coin, is the most disturbing activity of boosting the extremely high frequencies, so lazy cymbols gain the same dramatic impression as smashing glasses, way to fill the atmosphere with etherialy disturbing tweeps and meanies - gauranteed way to mess up a hippies trip.

Other than all that, it could be just a foible.


off-the-off-topic, back to the thread title

Fresh ideas, as always. Thank you very much.
Sure that's not transparency, but may well meet my taste as well. And to me it's not too far away from the idea of objective hifi. I will try it.
Edited:
"Do you happen to have details on the frequency response (amelioration, response drop beyond 14 kHz)?" -> Just tried. I see: frequency response starts going down at 9 kHz.
Sound is quite alright to me, but I'm not sure whether I will be content with such a rather extreme lowpassing. But your thoughts and practice are highly welcome to me, and I will go and make my own experience with it.

Added later:
Apart from normal music I tried trumpet and herding_calls from Gabriel's problem sample thead using 3.98a3 -V6 --vbr-new together with your kind of lowpassing. This kind of lowpassing reduces the artefacts significantly. So it's especially welcome when using low bitrate.
To me musical enjoyment is not really reduced. With bad recordings it may even improve (but I haven't found such a sample so far).
Makes me reconsider all my attitude towards high frequency behavior.

Thanks for the links. In the second thread is yet another link which points here where Gabriel says (emphasis mine):
This is the first time I heard that q1 was also affected.

You should also consider lowering the range of ca. 3-5kHz a little as this is 1) the most important range for speech intelligibility and 2) the frequency range where our hearing is most sensitive. But then again, the 15/2 fullrange boxes commonly used for smaller parties have this filter built-in.