I was just thinking, let's take a break from trying to make our MP3s sound as good as possible and instead try to make them as bad as possible, just as a fun little exercise.

The rules:
The file must average 128 kbps.
Must use a modern (3.90.3 or 3.96) version of LAME so that most of us can reproduce it outselves.

My contribution:

-k -q9 -m d -b128

Plus all of the following that are not already implicit in the above:
--noshort --nores --notemp --noath

Oddly enough, it still actually sounds decent compared to a lot of the stuff I've heard, especially on P2P networks and independent band web sites that sound like they were encoded by deaf people.

The NCTU guys came up with an "interesting" command line for lame in an attempt to make their AAC encoder "sound" better at an amusing test they conduced:

http://psplab.csie.nctu.edu.tw/projects/in...ac-v1.0rc1.html

They also messed up where they shouldn't in FAAC.

Heh, I just tried that commandline with 3.96.1 and it sounds a lot like a Bladeenc mp3 at 128kbps. At least to me. Similar annoying artifacts.

It also encoded insanely fast. 31x on a 2GHz Pentium 4.

Give this puppy a spin: -b 128 --noath --ns-bass -32 -X2 --allshort -k -q9 -m d --nores --strictly-enforce-ISO

Sounds a lot like Blade; nothing else I add will make it much worse. Actually, it sounds a lot like WMA @ 64kbps.....guess that M$ can now say that WMA sounds like a 128kbps file at half the bitrate. And they don't even have to upload a mono WMA file for comparison!

Edit: Wanna have some real fun @ 128kbps? Get your hands on an MP1 encoder like SoloH from ReallyRareWares. It will give you a new appreciation for WMA and ATRAC

You know. A worst command-line lame encoding job may not be such a bad idea to keep around. For a while, I had a business model in my head where bands could offer the world's crappiest mp3s of their music for free. And of course, if people like said music inspite of world's crappiest mp3 encoding, they could buy the CD/FLAC/MAC/etc.

It still might not be a bad model to follow.

umm... that'd just piss potential buyers off.

hmm. i like the idea of all short blocks with no lowpass - really lets you keep a "flat" frequency response

hmm... here's an idea: sample rate wasn't specified, so one could easily add a resample to 48000 into the mix.

and if you want, you could turn on --nspsytune and turn the midrange right down for some interesting effects.

No then you'd just end up with a new wave of bad mp3s over P2P networks. It's 128 Xing all over again :|

We've had this experiment already:

Is LAME capable of calamitous MP3 quality ?

I think we ended up with something like this:

lame.exe -V0 --preset radio -q9 -b256 -B256 --resample 48 --interch 1 -md -p --noshort --notemp --nores -k --strictly-enforce-ISO --nspsytune --athlower -56 --ns-bass 2 --ns-alto 12 --ns-treble 9

Edit: Removed redundant -k switch.

I just read that thread, copied that commandline and was dying laughing. Amazing how bad even LAME can sound. It takes real work to get LAME to sound bad. Some other encoders manage to do so with no help. 7h3y r 50 1337.

While mildly amusing, it's a little bit sad/frustrating that the frontend supports this kind of exploitation and misuse...

Most of the people here at HA laugh at these kinds of commandlines, while other people actually use them because they think they're doing the right thing..

LAME commandlines actually scared me away from it until the presets came about. Then I could feel comfortable with LAME. Typing --alt-preset standard is much easier than trying to form some huge commandline. It is funny to play around with it, but yeah misuse can lead to some really bad things.

To all newbies. These commandlines are a joke. They're just silly fun had by some people. Please do not archive your music with them.

AH! i forgot about the interch setting!

that really screws things up... (but is a good idea in principle, only that it should be limited strictly to MS frames only, and very small values at that)