when I burn mp3s to cd in audio cd format thingy does that uncompress them any or is it still mp3 quality?? I know dumb question but I'm new to mp3's

They are still compressed whether written as mp3's or converted to redbook. mp3's created at rates from 192kb to 320kb don't sound nearly as bad on cd as the smaller file sizes. Granted, they will take up more room, but will sound better. I like the free CD-Ex program for making mp3's. :yes:

There's no going backwards once you compress your audio in mp3 format. That's why it's important to select a higher bitrate when doing the compression and conversion from CD. 160 is good for average stuff...190 or better is necessary for high quality. 120 seems sorta low-fi to me...like FM radio...good, but lacking.

When you convert back to .wav and then to Redbook Audio (CD format), it doesn't 'expand back out'...it's only as good as the original source..."Garbage in, garbage out".

Thx I have some mp3's that sound really good but others are blah I assume the blah ones are less then 120. it was a dumb question but atleast I learned something :)

There's no going backwards once you compress your audio in mp3 format.

Absolutely correct. Once you rip the song from the source cd and turn it into an MP3 file, it's done. MP3 is a "lossy" encoding method, i.e. the file is made smaller by stripping audio data out. The logic in what data is taken out is very good, but the more you take out, the poorer the sound quality. No way around it. Burning it to a cd and then playing it doesn't put anything back.

As was mentioned in several posts, generally, 128 kbps or lower is not going to sound good on a decent playback system. 64, forget it. Although I've never personally tried it, I've been told that 128 kbps vbr (variable bit rate encoding) sounds pretty good.

If you want good sound quality, 192 is excellent, and 320 will get you music that sounds so good you will have a very hard time telling the difference between the ripped track and the source cd, even in a very good system.

Yeah, exactly!
I never download mp3s with less than 160 kbps. At 320 I agree it sounds much better, the extreme frequencies are still there.

In winamp you always see the files bitrate. Get the classic version of winamp though. (2.91 or earlier, the later editions are full of unnecessary crap)

Btw, is it a notable difference in getting a good soundcard? Just using integrated stuff on the mainboard. Maybe a E-MU series card would be alot better? (Supposed to have good DACs).

/Erik

A layman's description of how all compression on computer files works will lead to a better understanding of how this can be considered 'lossy' in an audio purist sense, but why at higher bitrates, it's unnoticeable.

When you strip down any computer file...audio, image, word doc, whatever, to the most basic level, it's all 0's and 1's. Every byte is a combination or pattern, if you will, of 0's and 1's. Say we're looking at a small square picture of a box, half of which is pure red, the other half, pure green. What we really have is a collection of bytes...maybe thousands of them...half of which have the pattern that makes up "red" and the other half which makes up "green". We can store these as thousands of red and green bytes, or we can store them as ONE red byte, and ONE green byte, and a few bytes of information on how many bytes to explode the one red and one green byte back out to, when we want to view it.

Same with audio compression to mp3 format. Similar frequencies and byte patterns are condensed as 'one byte' to store, making the file smaller. When playing back, everything gets blasted back out to it's proper place.

The bitrate is sort of a measure of 'how granular are we going to get with deciding what's similar and what's different'. Higher bitrate = more accuracy and less loss. Low bitrate = closer to the 'red - green' example.

I think that the 'average' mp3 file (at 120 kbps) runs at about a 10:1 compression ratio. That is to say, standard uncompressed .wav runs about 10 megs per stereo minute, and an mp3 at 120 would be about 1 meg per stereo minute.

There are many types of compression. One way to explain them is by using the buzzwords "lossy" and "lossless". Lossless compression implies that when uncompressed, the original data is retained in every respect - nothing, not even a one or a zero, is lost. Traditionally, this can only be done by using mathematical algorithms. An example - if I have a sentence with five spaces in it at a certain place, I could replace that with a symbol like '5s' that represents 5 spaces. On uncompression, the mathematics would see the '5s' and know that it means 5 spaces and put them back again. No loss. A 'zipped' file is an example of lossless compression. How much the file can be compressed is a function of how cool the math is that goes into it.

A "lossy" format compression is one that throws away certain information, with the idea that you probably won't notice the difference anyway. An example of this would be a jpeg photo file. If you compress a jpeg file enough, it no longer resembles the original file (maybe a bmp or tiff file or something like that). So a certain amount of lossy compression is ok, but too much and your image (or music) suffers for it.

Lossy compression can NEVER be put back the way it was originally. You can turn a ripped CD music file into a WAV file (lossless) and then compress it into a much smaller MP3 file (lossy), but you can't turn it back into the original WAV file again, ever. That missing information was thrown away and you can't get it back.

There is more to it than just that - but this is the basic concept.

Those who complain that CD's themselves don't sound as good as LP's have a valid point - and this has less to do with compression than it has to do with what is known as the sample rate. In an analog medium (LP's), there is no concept of ones and zeros like in the digital world. In the digital medium (CD's), everything has to be a one or a zero.

A quick-n-dirty analogy would be a volume control knob versus a digital volume LED. One turns from zero to max in an infinite number of steps. The other counts from 1 to 10 in steps of one. My TV has a digital volume control. Sometimes I find that volume 5 is too high - but volume 4 is too low. I can't do anything about that, I have to live with it. With an analog volume knob, I would set it between 4 and 5 at a setting I liked, no problem.

Now, take that example and apply it to the concept of time. On a moving line, music volume goes up and down (tone, tempo, timbre and everything else that describes music changes over time also, of course). A sample rate of 1 sample per second could catch the volume as it changed, but in a very rough manner. Probably would sound like crap. So make it sample thousands of times per second. You'd think that would be enough, right? Well, maybe. It turns out that the human ear may be more complex and discriminating than we thought originally - and perhaps can hear the difference when the sampling rate is lower or higher. That's why we now have DVD-A and SACD - the original CD sampling rate may be inadequate to represent what was actually in the original music to the extent that it fools the human ear.

MP3's then, take what may be already inadequate (CD's) and compress them EVEN MORE, in a lossy way (how much is lost can be controlled as others have mentioned, by choosing your encoding bitrate - 190, 320, etc).

Hope that helps. I'm just restating what others here have already said, hopefully in a useful way.

Best,

Wiggy

Cool, well put Wiggy.

So, in my example, "purple" in a lossy form of compression would be stored as "red", as it's closer to red than green. In lossless compression, "purple" would be "purple".

Safe to say that lossy compression applies to music and graphics, where file size can be huge and loss can be relatively unnoticed?

Whereas say, a Word doc missing the word "and" might not read so well.

So, in my example, "purple" in a lossy form of compression would be stored as "red", as it's closer to red than green. In lossless compression, "purple" would be "purple".

Correct, although technically, the colors don't change in jpeg compression. More like this - take a photo and reduce it to a grid, each intersection represents a point (a pixel) on that grid. Now throw away every third point. The picture probably still looks like it did before - the mind compensates for what is not there anymore. But there are fewer points, so the resulting file is smaller. If you throw away too many points, the picture suffers from 'jaggies' and 'stairstepping' and the mind rebels and refuses to see the photo as a photo anymore.

In reality, jpeg compression tries to be a bit more crafty than simply removing every third pixel - for example, it tries to remove more where there are a group of the same-color and less where the colors are different from pixel to pixel. That's why every jpeg file compresses to a different size, even though their height and width may be exactly the same. And you can adjust how aggressive the mathematics are that remove data - so you get smaller and smaller jpeg files, but they look less and less realistic.

Safe to say that lossy compression applies to music and graphics, where file size can be huge and loss can be relatively unnoticed?

Same thing for music, although psycho-acoustics may be even more complex and less-well understood than vision. Imagine this - if you fired a gun and hit a cymbal at the same time in the same room, do you think you would hear the symbol? Well, technically the sound is there for both of them - a microphone would pick up all the soundwaves, not just the loudest of them. But since we figure that humans won't know the difference if we remove that cymbol sound, we can probably do it and no one will notice. The result is space saved in the resulting file.

You can also do it by carving off the tops and the bottoms of the waves that sounds make. We say that humans can't hear sounds above a certain frequency or below another. Let's just say 20 hertz to 20,000 hertz for the sake of having a range (not everyone agrees what the range really is that people can hear). Now you can simply ignore any data that comes in over 20K or below 20H and supposedly, you've lost nothing in sound people can hear, although again the file size is reduced.

You can also compress the range - this is called dynamic compression and radio stations have to do it to broadcast an FM signal. That is, you apply a mathematical rule that reduces the difference between the highest and lowest points by an equal amount - the mind again compensates for what is not really there anymore, unless the dynamic range is really radically reduced.

And again, you can reduce how often you take a sample of the original sound - if you sample less frequently, you gather less data, and the resulting file size is smaller - but if you cut it down too much, the mind notices and refuses to play along with fooling you anymore.

Whereas say, a Word doc missing the word "and" might not read so well.

Correct again. The only difference between say, a Word document and a WAV format file is that the WAV file is sampled over time - the data is recorded at a certain number of times per second. So some data is lost regardless, but it is not intentionally thrown away like an intentionally lossy MP3 does. Since a Word document does not have a 'time' function, this does not matter to it.

The same thing, by the way, is what you see when you view video over the web - just like an audio MP3 file. You may notice that they ask you what kind of bandwidth you have. In order to stream data to you and not have it break up, they reduce it in a number of ways if you have a slow connection - they make the window you see smaller, they sample less frequently (jerky motion and sound), and they reduce the number of pixels (blocky images).

Best,

Wiggy