Older versions of LAME (quality thereof), Moved from MP3 - Tech Options

[Slitty]

I have a large 320kbps collection of mp3s from different sources. I've only recently began exploring the technical aspects of music compression, and subsequently noticed that parts of my collection were encoded with LAME as old as 3.88 and even 3.82. Should I expect these to be of worse quality than music encoded with newer versions of LAME? Is there any way to determine the version used of mp3s made with other encoders?

[halb27]

3.82 is really an old version of Lame, and there has certainly been progress since then. But luckily your files are encoded with 320 kbps, that's why I wouldn't worry.
EncSpot is a tool that tells you about the encoder used, but it does not always give the correct answer.

[Slitty]

Given some choice, what is the oldest version of LAME that you would consider on par with modern encodes?

[halb27]

As for CBR 320, 3.90 is a good ancient version.

[db1989]

Isn’t anyone going to mention 3.90.3? After all, it was an integral factor in the creation of Hydrogenaudio: before that, there was r3mix: the name of both a website and a preset (i.e. assemblage of tweaked parameters) to which Dibrom wished to provide alternatives, leading him to be a main developer of the --alt-presets, which were first introduced in said version as options and later assimilated into all operations of LAME.

[Wombat]

Hehe! 3.90 didn´t use -Z from the beginning so 3.90.3 must indeed be some "legendary" version.

[shadowking]

3.88 is decent for high bitrate CBR 192 + and GOGO is based on it. 3.90 is another good one allowing both the old and new methods. Really at 320k you have nothing to worry about unless some crazy switches were used.

[Canar]

Given some choice, what is the oldest version of LAME that you would consider on par with modern encodes?

If LAME improves every version, it's pretty difficult to give an answer to this, wouldn't you think? The earliest "dependable" VBR MP3 encoder, in my eyes, would be the 3.90.x series, but as db1989 says, I'm likely biased on the matter. At 320, things are quite different. You're probably well beyond the threshold of audibility in most cases.

This post has been edited by Canar: Mar 4 2012, 05:27

[Slitty]

Thanks for the information guys. Great stuff, I really appreciate you sharing your knowledge! I mostly wanted to know if I should replace some of my "older" encodings, even if for OCD purposes. It's comforting to know that it's no tragedy if I can't. But what about FhG encodings at 320kbps, do they hold up against LAME? Is there any way to determine how which version of FhG was used?

PS: one of my albums was encoded using something called NCODER, and even Google seems to have hardly heard of it! =D

[shadowking]

The only concern is an old encoder called BLADE which had terrible quality regardless of bitrate. FHG and lame are good unless you use a 90's version but even then 192 should be acceptable and 224 sufficient for harder cases. VBR was immature back then and can be worse than CBR. I remember joint stereo not being properly tuned with some encoders too so it is was often avoided for high bitrates. Gapless encoding appeared around lame 3.90 which is another bonus. Again for 320k I think you are safe.

[mjb2006]

Another thing to consider is that as our hearing worsens over the years, the quality of the older encoders improves.

[lvqcl]

unless some crazy switches were used.

for example, because of poor hardware:

I know [...] that I should use --alt-preset cbr 128 or --alt-preset 128 if I want an ~128 kbps encoding, but my crappy MP3 players skips and pops if I don't use the parameters below.
[...]
"--apeinput -b 128 -m s --nores --noshort --strictly-enforce-ISO"