I am planning to backup my CD collection to MP3 and have questions about LAME presets:
Is preset extreme generally worth the extra bitrate? How common are audio problems with standard that extreme can handle? (for that matter, does anyone think that -insane is a better option for day-to-day use on midrange hi-fi systems?)

Thanks for any comments or advice.

from my experiences, extreme produces slightly better results than standard, but this doesn't necessarily mean that you should go with extreme right away. do some blind listening tests and see for yourself. if you can't tell any difference between the two then you might as well go with standard.

just to add to my previous reply, in my opinion insane for "everyday" use is an overkill. just use either standard or extreme presets. if i recall correctly, the fundamental difference between the extreme and standard preset is that --aps uses a different ATH curve which allows it to keep the bitrate lower than extreme.

--ap insane produces better results on some samples: it reduces pre-echo on very sharp recordings (castanets for exemple), reduces distortions on instruments like harpsichord... I would say that the quality difference existing between --aps and --ape is the same existing between --ape and --api.

Thanks for your help, fellas

I have tried a few listening tests, and have not been able to detect anything but the tiniest differences between standard and extreme. Are there any particular genres or songs that I should be testing before I commit all my CDs to -standard?

Also, I would be interested in any advice from anybody with better ears than mine who has a strong opinion on this issue.

cheers

If you listen baroque music, especially harpsichord solo, then I would strongly suggest you to be careful with --alt-preset standard (and with MP3 format in general).

Could You please tell me what the situation is with MPC q5 with this type of music?

I'm not familiar with mpc --standard. Bitrate is higher than --preset standard (for solo harpsichord : 170-180 kbps for LAME, and 200-220 kbps for Musepack). Quality is much better, reaching transparency or being very close to that state. Some slight problems can be ABX with success, but it's nothing compared to distortions I heard with lame on the same music.

Guruboolez, do you listen Uriah Heep? What about their hammond? Does it mean I have to regrab & reencode all Uriah Heep albums in MPC or loseless? I sold all albums after I had grabbed and encoded them in APS 3.90.3. And what about Vorbis Ogg GT3 6.0 & 5.0?

What's that? Which century?

Since 1970 till now. Allmusic.com describes them as 'Album Rock, British Metal, Heavy Metal, Britain, Prog-Rock/Art Rock'
It's a pity you didn't listened them. Better I had not read this poll.
I must find out those CDs and do some tests now. However thank you Guruboolez for reply.

Twentieth

I think you are talking about completely different music.

Do you hear something strange when you listen to your encodings? Something annoying or simply odd, that would justify any ABX test? If not, then --alt-preset standard is fully transparent for you, and you don't have to bother about quality.

Placebo already did it's dirty work. Some of encodings seem to me distorted. But I can check them till I will find some CDs.

Only way for a complete piece of mind is to perform ABX comparison. Try to find a CD or lossless files.

Thanks

What is it about the harpsichord's sound that makes it difficult to encode as MP3? I have noticed (and ABXed) strange distortion with --aps in fairly simple acoustic guitar pieces (such as the first track from Friday Night in San Francisco by McLaughlin, DeLucia and Dimeola).

Is this a artifact of the MP3 encoding scheme itself? Or is it a LAME specific problem?

In about 99% of my collection, I can't reliably ABX --aps from the original CDs - so I am always very interested when I come across a sample that LAME has a problem with.

Try with a recent fastenc or other Fraunhofer MP3 encoder, at high bitrate. I've tried long time ago, and I'm not sure to remember correctly the results of the test. But I'm pretty sure that Fhg encodings were less distorted than --aps encodings.

Heavily loaded with high frequencies at a loud level :





...with transient-like spikes (one pixel = one sample ) :



Though the spikes are rather audible than visible.

So are any other codecs better for this kind of music? OR are future versions of LAME expected to handle this kind of stuff better?

To get back to the subject.
I think it comes down to your equipment.
The only time I notice the diff. between X3M and STD is with headphones.
I Find X3M to be much more "airy" and alive. I think it's because of the way X3M cuts of less of the high freq.s.

That's just my 5€ cents.
I'm sure no-one will take the words is such a low-poster seriously (I'm more of the listener than a speaker).

Actually I should think that using phrases like 'airy and alive' are more likey to get you taken less seriously than post count.

I don't think those phrases are ridiculous, but he didn't proved any ABX results - no objective evidence.
I think both objective and subjective results are good to combinate. But if doing only subjective test, it sucks IMO.
Think about if someone do ABX results, but he/she couldn't describe the different, it's better than just subjective test, but still not as good it should be IMHO ;)

Sorry, I didn't mean to imply that the phrase was ridiculous. It's just a bit remeniscent of what i've read on certain 'Audiophile' forums. You know, the ones where 'burning in' equipment for a week before listening is commonplace.

Ah yes, you have to do that... and be sure to buy only CD-Rs with yellow dye, because if you burn audio to them, they produce a more "punchy" sound