Best parameters for Lame 3.92?

As my DVD player dont recognize mp3 files with VBR, I had to fix the compressíon.

These are the parameters I’m using:

-b 192 -h %S %D

Besides less compression, are these the best parameters for a good audio quality?

“–alt-preset cbr 192” is the best if you want CBR @ 192 kbps.

For GOOD audio quality in CBR mode you would have to use “–alt-preset cbr 224 %S %D” at least. 192 kbps is NOT enough for good audio quality :Z

aaah, jesus christ, you mean I’ve been making crappy 192kbps MP3s when I needed at least 224? But good to know. Thanks.

BTW, what does %S %D stand for?

You see everyone has different ears - and hence what seems transparent to you may not be transparent to someone else. So its you who has to decide what bitrate suits you best. 192 kbps is a general ‘theoretically safe’ level and gives you good quality at decent sizes.

If you want to be theoretically perfect, then go in for alt-preset-insane - that gives you CBR 320kbps. However if you find 192 kbps transparent, go ahead with it.

And yes, I suppose %S=Source and %D=Destination

Originally posted by ExpertTech
[B]You see everyone has different ears - and hence what seems transparent to you may not be transparent to someone else. So its you who has to decide what bitrate suits you best. 192 kbps is a general ‘theoretically safe’ level and gives you good quality at decent sizes.

If you want to be theoretically perfect, then go in for alt-preset-insane - that gives you CBR 320kbps. However if you find 192 kbps transparent, go ahead with it.

And yes, I suppose %S=Source and %D=Destination [/B]

If it were perfect it should be CBR “infinite” kbps :slight_smile:

I wonder if people could hear the difference between CBR 320 and CBR 9999999999999 :slight_smile:

Originally posted by Stoner
aaah, jesus christ, you mean I’ve been making crappy 192kbps MP3s when I needed at least 224?

There are two problems.

The first one - a lot of sounds need more than 192 or 224 kbps. Usually they are short sounds, but they often occur many times in one second of audio. With VBR there’s no problem - encoder increases bitrate for a fraction of second. The file is not too big and sounds fine if you use use the right encoder and right setting. With CBR, altough there is a bit reservoir, it’s usually not enough to compensate for limited bitrate. So 224 kbps is the bottom line for GOOD audio quality. 192 kbps is just not enough.

The second problem is that MP3 is a poorly (at the moment) designed format. It has gigantic problems with compressing very short sounds and frequencies above 16 kHz. Usually it can’t store high tones not to lower the quality of the rest of the sounds. So, the higher the bitrate, the better MP3 sounds and there’s no end to this line. :frowning: Dibrom from Hydrogen Audio has designed alt-presets and they do a really good job, but that’s virtually everything one can expect from MP3. The limitations of the format itself cause, that it is never going to sound perfect. Very good - yes, in a lot of cases. But usually - not perfect.

So, don’t even think about going lower than 224 kbps, if you can’t use --alt-preset standard, because you won’t get good quality. :frowning: Sad, but true.

actually, as has been posted before, there are graphs of cbr recordings, and how close the frequencies they hold are to the initial wav file. pretty interesting, but at 192, the mp3 starts to stray somwhere between 18khz and 21khz, if i do remember. if your ears are gonna miss those kind of highs, you’ve got issues. no offense meant :wink: … of course, the ‘busier’ the music is, the more samples need to be taken in order to preserve the original quality.

If you think that you can tell the sound quality just by looking at the spectrum of the encoded signal, you are disastrously mistaken, I’m afraid.

Graphs tell very little about sound quality. Very, very little. I can show you the graph from 128 kbps MP3 which looks better than the graph from 320 kbps. Do you know what I mean ?

The amount of certain frequencies present on the graph tell nothing. If you compared the graph of the wave compressed with non_existing_ultimate_encoder, and the graph from normal encoder, the former would look much worse.
Psychoacoustic encoder has one basic goal : remove as much as possible without making these changes audible. So it would remove BIGGER portion of the signal while still maintaining the same quality. The graph would look much worse, the sound - would be exactly the same as in wave.

So, if you want to compare the sound quality of two files, one with better and one with worse graph, get the original, download PC-ABX from http://www.pcabx.com and check, which one is harder to discern from the original. You could tell that Musepack sonds worse than MP3, because it has worse frequency analysis graph, but you would be really surprised by the results of the test, I think.

you cannot show me a graph of an mp3 at 128 that looks better than the graph of one at 320. it defeats the entire purpose of encoding into 320 in the first place. the graph would show you the frequencies that were included in each frame of the sample - and obviously, the 320 encoding can handle more frequencies, as it has more space per frame.

didnt you just argue 2 posts ago “dont think about going lower than 224 kbps […] because you won’t get good quality” ?

and of course, the graph doesnt tell the “amount” of frequencies present, rather, it shows the range of frequencies present. its not like the graph says “well, i have 30 of 100 hz, and 55 of 120 hz”. what sense does that make?

sorry to be so doubting, but i just have a problem agreeing with most everything you posted. sorry again.

Well, I’m sorry, too. I haven’t made my point clear enough. Here is what I mean, I hope this version will be clearer.

You cannot judge the quality of a wave, which has been psychoacoustically compressed, by just looking at the graphs. The graphs alone tell something about the file, but you cannot say, that some setting, which leaves more high frequencies in your file, is better, than the other, which leaves less of them. Here is the example :

–alt-preset cbr 256 supplies frequencies up to about 20.5 kHz. But, I could use --alt-preset cbr 320 --lowpass 18.5 and it would actually give worse looking graph, but better sounding file. Hearing 20.5 kHz tone alone is possible, but hearing such frequencies inside the actual music, where they are being masked by lower, much more audible frequencies, is almost impossible. So, the space inside 256 kbps file would be just wasted in most cases. So, you would have better looking graph and worse sounding file.
Take Musepack as an example. It doesn’t use lowpass at all, but if you look at the graph, you will see much less of high frequencies than in 320 kbps MP3. Yet, Musepack sounds much better.
Graphs usuall tell nothing about the quality itself. They just show you, which frequencies are present, but sound quality is a totally different story. So, if you see high frequencies inside 192 kbps file, it doesn’t mean it sounds good.

MP3 format lacks the scalefactor for the last scalefactor band, which holds frequencies above 16 kHz. Because of that, high frequencies are being encoded to much bigger size, than lower frequencies. In metal music, for example, there are a lot of audible high frequencies, so the encoder shouldn’t leave them out. But keeping them is a tradeoff between them and lower, more audible frequencies. But, if those highs are audible, they should be stored in the file. With 192 kbps you just doesn’t have enough space not to compromise quality of those lower frequencies, when storing these highs. It’s a limitation of the format itself - poor design. The whole point of using 224 kbps is just the maintaining of resonable sound quality, when there’s no space left in 192 kbps frames.
Like I said, a lot of sounds require more than 192 kbps. In fact, a lot require 256 and above. Yes, because of the issue with missing scalefactor, too, but not only because of that.

If you want to test it, compress something with lame --alt-preset cbr 192, then with --alt-preset cbr 256, decode both files to waves and use PCABX to make a blind test. Compare this way 192 kbps with original, then 256 kbps with original. If you use some metal music, you will hear, that with MP3, 192 kbps is not enough. :eek:

I hope it’s clearer now. Sorry about so fuzzy post.

Well, tried EAC w/ Lame 3.92 last night. I entered the parameters, expecting CBR 224bit files, but got a whole buncha VBR files. They were secured (whatever that means), but they don’t sound as nice as rips I did w/ audio catalyst @ 192bit. Don’t think this EAC w/ Lame 3.92 thing is for me.

Instead of entering parameters within EAC why don’t you rip to WAVs and then use the command line on the ripped wavs? This shoudl definitely work!