I was wondering if some one could help me here. I'm interested in making an MP3 with all the artifacts but full band width. I'd really like to have that 'underwater' sound.

Or even better. Remember before mp3 got really big? Sometimes you would download a file and it just screetch and sound garbled. That'd be brilliant if I could do that!

Tony

IIRC there was a thread called "How to make mp3s of calamitous
quality" in the mp3 or off-topic section... Try the "search" feature.

I searched a few times but still can't see anything. I either get no results or hundreds :-(

Got it!

http://www.hydrogenaudio.org/forums/index....&hl=mp3+quality

and

http://www.hydrogenaudio.org/forums/index.php?showtopic=4457

I think a lot of those switches are redundant. For example, any switches related to the psymodel are irrelevant when you use -q9 and -k, which disable the psymodel.

A litte bit of experimentation on my part has given me this:

--vbr-new -b128 -B128 -q9 --resample 48 -md -p --noshort --notemp --nores -k --strictly-enforce-ISO

For best results, try on a sample with a lot of high frequency content.

The sad part about this is that despite what this sounds like, I have heard worse. I have heard burned compilation CDs that you would have to be literally *deaf* to think that they were acceptable quality. And I'm not talking about acceptable quality by HA standards, that's way above even most enlightened listeners. I'm talking about completely ruined sound on a level that you don't even have to be attuned to MP3 artifacts to hear, that the average "CD quality 64 kbps WMA rip" user would listen to and say "that sounds bad."

Even with that, the VBR and -q9 switch is redudant. Here is what I got when I started encoding a file:



So maybe if you use CBR as opposed to VBR you would get a lower quality? Or it might be my version of LAME though.. I used 3.90.3 with this.

Just encode a tune with lots of high freqs with Blade at 96kbps. Or even 128.

The intro to Hotel California messes Blade up too.

I think he wants terrible quality and high bitrate at the same time. Maybe someone could make a specialized version of LAME for this, but it would have to be kept private to prevent just any idiot from thinking it will suffice for proper archiving.

Hehe, that would be very interesting.. have it have like have different crappyness settings

Strangely enough, CBR actually sounds better than VBR in this case. I think it's that the psymodel is assuming that it can use variable frame sizes, but you're constraining it to a single frame size with no bit reservoir so it gets messed up. In any case, replace -q9 with -f to get rid of error message. According to the switch reference, -q7 uses the psycho model only for pre-echo and M/S, which makes it moot since we're using all short blocks and dual stereo.

I've changed my mind about --noshort. --allshort sounds much "better". You don't get pre-echo anymore, but that's a small price to pay for TOTAL AUDIO ANNIHILATION, MUWHAHAHAHA.

That commandline again:

--vbr-new -b128 -B128 -f --resample 48 -md -p --allshort --notemp --nores -k --strictly-enforce-ISO

Now that was extremely painful! I just tried it on Dancing Queen by ABBA, the perfect song for this kind of thing hehe.

Why don't you try with Blade? http://bladeenc.mp3.no/skeleton/DL.html
Old Xing could compete but thats not free

I want to make an MP3 with some inferior MP3 encoder with no-insane command line just like simple cbr 128 or 160...
Are you suggesting me Blade & Xing (old)?
Are there any other not so good MP3 encoders that sound badly on simple cbr 128 /160 Kbps?
I've found on www.ff123.net that it's not so black & white with the inferior quality of Xing encoder anymore...
Thanx a lot in advance!

PS: I want to show my friends what's the difference with using LAME --alt-preset cbr and other --alt-presets...

One word, transcoding. That's how the experts at the file sharing networks get artifacts in their music.

or check out Really Rare Wares for the best worst mp3 encoders.

Fine. Green Day Basket Case at Blade 128. This is unmistakeable shit compared to alt preset 128.

Really, if you play with blade at 128 with complex music it will become very obvious with the artifacts. Not close call abx stuff. Just shit hitting the fan.

Blade is not that bad even at 128 kbps. I have done ABX tests on the the waiting sample & the only thing LAME 3.90.3 has over Blade in this sample is the alt-preset system. I tested LAME: 128 kbps, high quality, Blade: 128 kbps, default settings & was able to ABX both up to 25 trials & LAME against Blade up to 10, LAME sounded slightly worse than Blade.

I also tried -alt-preset 128 & wasn't able to ABX it at all.

What is the waiting sample?

http://ff123.net/samples/Waiting.flac

Try different samples. I don't doubt that in certain circumstances Blade will perform well. Clearly you cannot be suggesting that Blade at 128 is superior to Lame abr 128 on a wider variety of samples.

Generally it is the worst of all encoders and displays many of the artifacts that the thread starter wishes to demonstrate.

The itunes encoder has been tested as very poor as well but I have no personal experience with this.

Blade on the other hand I have played with and I believe the two samples I have mentioned are fairly representative of its performance on many difficult clips.

Well have you tried ABXing LAME (without presets, & VBR) vs. Blade at the 128-256 range on a wide variety of samples? you'll probably find it's just as bad

The worst mp3 encoders are of course the first ones, try to find early LAME releases or Blade, what ever

blade at 320 will even give you bad quality although
it will only sound very cut-off (lowpassed)

try one of the first xing encoders.
now that at 128 will give you bad quality (yuk)

bc&peca

Blade at 320 is ok for a lot of "easy samples", I guess it sounds like Lame 192 CBR, but if you try some "hard-to-encode" samples, with a lot of impulses (drums, castanets...) , you'll hear HUGE pre-echos.