How much is lost when converting to MP3s?

G

GreenDank

New member
I've got some questions about MP3 conversions....this mainly stems from the fact my songs are sounding killer on my computer but like dog crap when I convert them to MP3s and play them in my car. HELP

Is there really much loss when converting WAVs to MP3s? I can't tell if I'm hearing the difference, or if it's just my crappy mastering that is sounding bad. I swear once I do the conversion my recordings lose a lot of umph. But when I convert songs from CDs they don't seem to lose as much. Why is that, just that the CDs are better produced?

Let's say you're working in Cool Edit, and you save it as a WAV file, then you convert it afterwards to an MP3. Is that different that saving the file as an MP3 in Cool Edit? Are certain programs better at converting, or are they all doing the same thing?

Are there mastering techniques to use when you know the end product will mostly be played as an MP3 (certain compression ratio or EQ setting), kind of like how songs for the radio are mastered differently sometimes?
 
part of mp3 coversion is dropping out all of the frequences that humans can't "hear", and a little extra too. I think they cut everything above 15kHz but I'm not sure exactly. yes, different converters use different algorithms that sound different. I use bladeenc, it seems to have the best quality I have found. what is always a dead giveaway of a cd burned from mp3s is that the cymbals sound underwater. most people can't really tell though. they also cut the lows, probably below 50 hz, so that might be some of the loss of "power". better produced mixes can take more abuse without falling apart.

what you are hearing in your car may or may not be related to this. if you download blade (its free) and still have these problems (make sure you are converting to at least 128 kbps) I would suspect that your mixes are just not translating properly.
 
you mentioned a program called blade? is that the best free mp3 converter you know of? if not, what should i get?
 
I'm not sure about this, but I think due to licensing issues that the Blade encoder wrapper is no longer offered (somebody correct me if I'm mistaken aboutt that.)

But even if Blade is no longer available, that's not really an issue. Blade used a version of a common encoding algorithm loosely called the ISO Codec. It is a fairly common codec used by many of the freeware and shareware encoders.

While ISO is a good codec, most (but admittedly not all) engineers perfer the Fraunhoffer (sp?) codec. This algorithm is the one developed by the Fraunhoffer Institute which was instrumental in the development of the original MPEG Layer 3 (now commonly known as "MP3") encoding engineering design. Most pro MP3 encoder plugs used by the companies who manufacture DAW software use the Fraunhoffer codec.

The drawback to the Fraunhoffer codec is that it is closely licensed by the Institute, which is why the MP3 encoder in many pro audio editors is not free. If you find a freeware MP3 encoder that is based on the Fraunhoffer algorithm, it is not legal.

All that said, MP3 takes a suprisingly large amount of information out of the music, even when encoding at high bitrates. For example, at the medium bitrate of 128K, a stereo MP3 file consumes approximately 1MB of disk space per minute of playing time. A mono WAV file, on the other hand, consumes approximately 10MB per minute. That's a 10:1 compression ratio, meaning that about 90% of the information is lost or synthesized when put through an MP3 encoding. And that's also a comparison of mono WAV to stereo MP3. Compare to a stereo WAV and the numbers go a bit higher still.

It's not just frequencies you can't hear like ultrasonics; much of it is stuff well in the range of human hearing. What the compression codec does is mathematically subtract sounds that by a psychoacoustic theory developed at the Frau. Inst. "don't matter that much" to the perception of the average human ear. This theory works fine for those listening passively via their boom boxes or iPod earbuds, but as anyone with halfway decent gear and active ears can attest, there is a large difference between the sound of a source WAV and it's MP3 encoding.

If you want to hear exactly what's missing from your MP3, take a WAV track, and encode a copy in your best MP3 encoder. Take the WAV and MP3 and import them both into your favorite editor as track 1 (WAV) and track 2 (MP3). Phase invert track 2 and then mix the two tracks together. The resulting mixdown will be all the sound that was in the WAV file that is not reproduced in the MP3. You'll be amazed at just how audible and recognizable the missing information is; you'll hear a lower fidelity rendition of the song or track. All that is being removed by the MP3 encoding process.

Class is over :). All that I just said notwithstanding, I'd hazard a wild guess that the MP3 conversion itself is not your problem. More often than not the problem is that the mixes just aren't translating to the "real world" playback systems that well. This could be due to the monitors your using for mixing, your room acoustics, or both.

G.
 
SouthSIDE Glen said:
If you want to hear exactly what's missing from your MP3, take a WAV track, and encode a copy in your best MP3 encoder. Take the WAV and MP3 and import them both into your favorite editor as track 1 (WAV) and track 2 (MP3). Phase invert track 2 and then mix the two tracks together. The resulting mixdown will be all the sound that was in the WAV file that is not reproduced in the MP3. You'll be amazed at just how audible and recognizable the missing information is; you'll hear a lower fidelity rendition of the song or track. All that is being removed by the MP3 encoding process.
Click to expand...

I've done this with a System of a Down track (lots of distorted guitar = lots of overtones). I also did the same thing with an OGG Vorbis compressed song (same song). The results amazed me. The MP3 tossed a LOT of higher frequency information, while the OGG tossed so much less it was hardly noticable at all.

If anyone is interested, I still have the test files and I could upload them if you want.
 
hey, southsideG that was one of your best threads imo..
great sht and a DIY test to boot.
 
Last edited:
SouthSIDE Glen said:
A mono WAV file, on the other hand, consumes approximately 10MB per minute. .
Click to expand...

Great post but think you have a typo. its actually a "stereo" WAV that is 10megs/minute if you are talking about 16/44.1
 
Ronan said:
Great post but think you have a typo. its actually a "stereo" WAV that is 10megs/minute if you are talking about 16/44.1
Click to expand...
Yeah, you're right, Ronan. A mono WAV runs half that at about ~5MB/min. I should have compared apples to apples, stereo to stereo, but I got turned around.

One of these days I'm going to learn not to write three-mile-long stream-of-thought posts without heavy proofreading first. Thanks for the correction.

EDIT: I wanted to go back and correct the language in that post to correct the stereo/mono error, but for some reason that is the only post of mine that is not displaying an "Edit" icon, so I can't edit it. My stupid mistake is stuck in amber. :(

G.
 
Last edited:
treymonfauntre said:
just don't encode anything less than 192. higher than that is better, obviously
Click to expand...
Better, but not by much, really. Go ahead an run that test in my previous post using your best encoder at, say, a 320K bitrate. The stuff taken out of the WAV file will still be disgustingly recognizable. MP3 is simply the worst thing to happen to music playback quality since the 8-track tape cartridge.

G.
 
I hate the fact that Ogg Vorbis didn't get picked up by the big companies that make mp3 players. It's so much better than mp3, yet there really aren't that many players that can play them. Even wma is more popular. It's a shame.
 
Halion said:
I hate the fact that Ogg Vorbis didn't get picked up by the big companies that make mp3 players. It's so much better than mp3, yet there really aren't that many players that can play them. Even wma is more popular. It's a shame.
Click to expand...
The quality is definitely there with OGG, but it's an independant standard. "MP3" is actually shorthand for the "MPEG 1 Layer 3" specification advanced by the MPEG standards comittee, and is actually a rather old standard. Before "MP3" was anything close to a household name, the MPEG Layer 3 spec was already being used by Macromedia as the core of their SWA streaming audio standard, which at the time was far superior in sound quality to the competition such as Real Audio and other similar technologies at the time. By the time OGG had roots, Layer 3 was already pretty well settled upon in the labs.

What I'm waiting for is for a new lossless standard to be adopted that can be edited in real-time with an invisible conversion stage so folks can excahange compressed files without having to convert. Whether anybody important is actually working on that now I don't know.

G.
 
Last edited:
GreenDank, why are you even converting them to MP3 first? You're adding another step which will make the quality worse.
All burning applications SHOULD be able to burn WAV/AIFF files.
 
Also there is a good amount of evidence that ultrsonic information, while it's inaudible to human ears, does affect how we hear the stuff that is in the audible range. So if MP3's cut off the highest freq's, then it's gonna affect the overall sound in a negative way.
 
Again, MP3s cut off much more than just the highest frequencies. The results of summing a WAV and an inverted MP3 are extremely audible across a fairly wide frequency range.

G.
 
SouthSIDE Glen said:
Again, MP3s cut off much more than just the highest frequencies. G.
Click to expand...
I didn't say it didn't. I was just pointing out that many people say that it doesn't matter if you cut off the ultrasonic freq's because they're inaudible but they do have effects in the audible range.
Personally I've never understood how anyone could say that MP3s sound as good as wave files .... the difference is clear. If you can't hear a difference then you're due to give your ears some more training. Even high bit-rate MP3s have a loss of 'air' and especially the sound of 'space' which is affected by reverb tails and such getting cut off.
For that matter ..... 16bit/44.1k CDs cut off reverb tails and other subtleties which is why there are quite a few of us that find them unsatisfactory.
 
Alexbt said:
GreenDank, why are you even converting them to MP3 first? You're adding another step which will make the quality worse.
All burning applications SHOULD be able to burn WAV/AIFF files.
Click to expand...

If I'm going to upload an MP3 to my site, and here when I'm getting feedback, I was just saving my mixdowns right to MP3 instead of as WAVs that I'd have to convert to MP3s anyway. For the real deal, I will take everyone's advice and not do that...
 
You must log in or register to reply here.
Back
Top