I like to pose a question for those who have done some listening tests.

Fisrt, I read here - http://www.audio-illumination.org/forums/i...5bf5b6d4d7bbf0d
That --extreme is slightly better quality then --standard.

Then,

I read these things here - http://www.audio-illumination.org/forums/i...897cba3379f42c0
1) That --extreme can fail in most the same cases that --standard does.
2) And --standard was designed to be as transparent as possible.

I also read these here - http://www.audio-illumination.org/forums/i...897cba3379f42c0
1) That --standard does have some problem files... Which seems to confirm the above.
2) That --extreme encounters that same problems... which also confirms the above, and....
3) Using alternate solutions like '-Z' or '--ns-bass -8' was a better solution in those, and maybe other cases.


So, I would like to pose these questions for people with experience -

1) On average, how many problem files (say per 100 CDs or so), did you encounter using --standard, and how many of those problems could be fixed by using --extreme.
2) And how many of those problems would '-Z','--ns-bass -8', or another switch fix.
3) And do you still consider --extreme better then --standard. ? (or worth it considering files size etc...) ?
4) And, finally, on a curious note, which was developed first... --standard, or --extreme... ?

Thanks for reading....

There are problem cases and "problem" cases. If a clip is not transparent, it's not automatically regarded as problem case, because we have to take into account the limitations of mp3 format in general and limitations of the current psymodels.

Anyway, -Z fixes most of the known problem cases for both presets. Those cases are regarded very rare and are the results of the same things (inaccurate use of certain noiseshaping type). No other similar cases have been presented, although there no doubt are those. But at the same time, you have to remember that lossy audio is never completely perfect, never totally transparent.

Of course extreme is at least theoretically better: it uses for example more sensitive hearing curve (athtype 2) and higher lowpass.

Standard was developed first when the internal code-level tweaks were developed, extreme was developed by adding a bit extra quality from the base of standard.

If you are concerned about the -Z issue, but also want to reduce the bitrate increase effect by the use of -Z, try --alt-preset standard -Z -Y. It will keep the bitrate "under control" pretty nicely, although over 16khz freq response is not so accurate anymore.
Dibrom will say that it's the wisest to use just --alt-preset standard, because the problems are regarded too rare.

Hi, thanks, i'm a bit more clear on things now....

However, i am amused that no one wants to commit to saying that '--extreme DOES sound better then --standard'. It seems that everything i've read so far states that 'in theory' --extreme should sounds better, but no one so far that i found has admitted to actually hearing a difference. (i'm sure i couldn't)...

As for the -Z-Y, i think i might stay away from messing around with them. For a couple reasons ->

1) You've stated, and i've read, that the problems are 'very rare' and 'too rare' to give much consideration too... I'm not sure if it would be productive to batch encode ALL my files with -Z just to try to eliminate the 'rare' problem case. (or would it ? How much does the file size increase with '-Z' ? And does it take away from the quality of 'non-problem' files ?). I would still like to get an estimate on how often a 'problem' track occur too... just to be preparded...

And 2) I don't think i want to loose the quality of the +16kHz range.... I haven't had my hearing tested, but i believe most people can still hear btween 16kHz and 20kHz... am i right on that one?

Anyways, i gotta run right now, but i'd love to get a responce.... especially to the 'can anyone acutally hear the difference' question.... .... Double thanks

I think the best audio coder is a compromise between all encoded audio clips.. because, the extreme clips does not represent the majority of audio clips. Tuning to extreme clips may cause the encoder to perform badly on ordinary audio clips.
Remember that Psychoacoustic principals are based on statistical results and therefore there is no E=MC^2 type of equation in psychoacoustic..

wkw

I've yet to hear anything odd in my 10+gig 'daily' playlist of --alt-preset standard mp3s, so I'd agree that --alt-preset extreme isn't really necessary. As has been stated quite a bit, 'standard' is meant to be the one to use for everyone and everything, and rarely has problems with any music. You shouldn't have much trouble hearing things over 16KHz for sure, yeah. For test signals my hearing falls off somewhere in the 18KHz range. I'm 31 and have been to many very loud concerts, dozens and dozens, many of them metal bands. I've got a stereo in my car that's pretty loud. So my hearing's been pretty abused over the years, yet I still have a pretty decent frequency hearing ceiling, so it's probably pretty safe to say that most people should too.

It depends on:

- ability to listen and history of damage done to your ears..........

- equipment:

- Listening from HD, with noisy PC in same room, is not a HiFi listening environment.

- Headphones:
with low budget (150 € eg.) you get nearly perfect speakers, you need a CD- player or DVD/MP3-player and amp, or good soundcard and silent PC.
perfect regarding frequency reproduction.
not so perfect regarding wide stereo image.

- main speakers, amp/receiver, DVD/CD/MP3-player or PCsoundcard, digitally to amp
perfect regarding stereo image, if room is appropriate, big enough
frequency reproduction: if the sound shall be really high quality, you have to pay for the speakers........


So, you (btw me, too) could be satisfied with 128 kbit/s MP3 by Car-Hifi/portable player.
But for other usages don't mind to double the bitrate. But it is not necessary. standard settings in Lame or mpc are well.

Actually tuning of a lossy encoder should be done so that every kind of music/signals is taken into account. It's also not good if extreme clips are totally forgotten, this is because there are whole genres of music which can be very much considered as "extreme" regarding lossy audio performance..
Best is to use as wide variety of music/signals as possible, and try to come up with the best compromise.

If you didn't notice, none of the standard's profile listed "problem samples" are traditional extreme clips, actually very far from it. They are just special cases, but not "extreme" like fatboy,castanets or death2. Even though standard is not absolutely transparent with extreme clips, there's no point to list those as problem cases, for one, because everybody knows them, and they are very much stretching the limits of what is possible with mp3, especially when you try to keep the bitrate in control. At the same time, these extreme cases are taken into account, like they should be, but not tweaked only according to them.

Size with -Z will increase considerably, depending heavily on the case. It shouldn't make the quality worse, just the opposite.

Many people who can hear up to 19-21khz sine signals, have trouble to hear difference with actual music between about 18.5khz and non-lowpassed. -Y switch is not a lowpass, but only allows clearly higher energy certainly audible over 16khz frequencies to be included.

Can anybody hear a difference between standard and extreme.. certainly sometimes, but constantly? Not necessarely, and even when there's a difference, it's often most likely not very remarkable.

Dear JohnV,

You said -

Actually, i'm not aware of what would be considered "problem" or "extreme" cases and genres of music.... could you list or direct me to where i could find what might cause me problems while encoding.... I've decided to use --a-p standard on all my CDs, but i think you listed "fatboy, castanets and death2" as possible problems.... so i'm going to use --a-p extreme on my Fatboy. And i was wondering what else i should set aside for extreme....