I'm wondering what people think is better here. I was pleased when I learned of -preset medium, because for my listening needs -preset standard is larger than I want to go, but I like having VBR that actually adjusts to the difficulty of the music, by song and section of song.

I've searched the forums and have found precious little actual evaluation of -preset medium. Mostly just thoughts and opinions. I've read a couple of times that users might be better off encoding to -preset 160 (ABR) over -preset medium. The thought is that ABR is better tuned than medium.

So I'm asking: do people have any sense from experience as to how -preset medium actually performs? Is it better, worse, or comprable with -preset 160? I'm not a golden-eared listener, nor do I really have time (as a Ph.D. student - and yes, it's going well, thanks for asking) to muck around in testing. And there are those with better knowledge than I here, anyway.

Note: I encode medium at -preset medium -b 128, based on all of the comments I've seen talking about how LAME 3.90.3 has a minimum bitrate of 128 that saves -preset standard's transparency on various screwy samples where the psymodel fails.

Thanks much.

preset (bitrate) is simply ABR for a given bitrate. One advantage is filesize is predictable. Preset (medium, standard, ext) are tunings for different quality levels:
near transparent, transparent, transparent + headroom. Bitrare is harder to predict but these presets are tuned for quality. I don't know if ABR 160 is better than preset medium. Personaly I would stick to the VBR presets unless you need the bitrate controlled.

Also I suggest not using -b128 or any switches - the presets are already tuned and you generaly can't make them better. You might even loose quality.

Right, right, right. But this is theoretical information. I don't need the bitrate controlled more than just in a general range. But again, I'm wondering, since I've heard people question how well tuned -preset medium is, how it compares to -preset 160 in their experience.



Good point. Though my understanding of -preset medium in Lame 3.90.3 was that the -b 128 was basically tacked on into the code internally, I think some later versions of LAME (not yet recommended) don't have the hard-wired 128 floor.

Again, the first question in this post is my main interest. The -b 128 floor is interesting, but it's secondary. Thanks again.

I'm not sure to the extent that this holds true at medium bitrates, but it has been said that VBR is generally better at handling pre-echo than ABR or CBR. If that is the case, --preset medium might be better on some samples (like castanets) than --preset 160. In real-life listening, I'm not sure that there is much of a difference between the two; in the past, I used both for a portable and neither struck me as being more or less transparent than the other. Of course, only a double-blind, placebo-controlled listening test will provide real answers.

Bitrates for medium generally ranged from ~134kbps up to about 190kbps on material I encoded, averaging 160 or so. On some quieter, older CDs, you may save a trivial amount of space with --preset medium over --preset 160. Again, whether or not the quality is the same has yet to be verified. You may also want to try --preset standard -Y, which is generally transparent to most people (frequencies >16Khz aren't crucial or even missed in most music), and results in bitrates of 160-175kbps. If not completely transparent, --preset standard -Y will at least be very damn close, which is more reassuring than using the medium or ABR presets in this neighborhood, and for little extra space.

fyi, the most informative thread on this topic:

http://www.hydrogenaudio.org/forums/index....showtopic=19350

I've not found too much about what the LAME developers think of medium.