View Single Post
  #34  
Old 2007-12-13, 01:06 AM
kupietz kupietz is offline
 
Join Date: Mar 2005
Lightbulb Re: mp3/lossless question for audiophiles...

I can most definitely hear the difference between low-bitrate MP3 and lossless, even on $5 speakers. You have to know what to listen for. You know when you get a recording with a really well-defined stereo image, where you can hear precisely where each instrument is placed from left to right? At 128k, MP3 "muddies" that. Things are harder to place in terms of position, because joint-stereo MP3 encoding moves them around in the stereo soundstage. Still, in my own subjective tests, I can't tell the difference between higher-bitrate MP3 and lossless with just my ears.

But here's a test you can do if you really want to objectively hear the difference is with your own ears.

Get a CD with a lot of energy & dynamics. I used "Making Plans For Nigel" by XTC. Rip it to a WAV file. Then, encode the WAV file to MP3 using the highest-quality conversion settings you can. Then, use whatever audio program you've got to expand the MP3 *back* into a new WAV file.

Then, go into your audio editor and open the new, lossy-sourced WAV file. Use your audio editor's "Invert" function to turn the entire waveform upside-down. Then, open your original (non-MP3-encoded) WAV file and use your editing program's Paste function to "add" it to the inverted one - mixing the two waveforms at the same volume to create a new waveform. This way, all audio that is present in both the original and the MP3-encoded file is cancelled out by this process - leaving you only the sound that is present in the original file but not the lossy one.

So when you play this inverted-and-added file, what you hear is everything that MP3 encoding removed from the original file. You'll hear some highhat sizzle, the top end of the guitar, maybe the snare, maybe some crunge off the vocal as well. It's an interesting experiment.

Using this test, I was able to determine that LAME VBR -V3 provided the best tradeoff between filesize and sound quality for listening on my proudly non-iPod MP3 player. Using V2 or V1 increased the filesize didn't significantly reduce the amount of sound I heard in the added-and-inverted file, and using -V4 significantly increased the sound in this file - IE, significantly increased the amount of audio that the lossy encoding removed form the original file.

Using CBR is, as Monty Python used to say, right out.

Fun fun fun.

Another thing you can do to hear the encoding loss is take an audio file and subjected it to several generations of MP3 encoding... convert it to MP3, back to WAV, back to MP3, back to WAV... do this 5 or 10 times and it will greatly exaggerate the lossy artifacts, so you'll get an idea of what even one pass of encoding is doing, ever so slightly.

What's interesting is that people are so used to tape, they assume "lossyness" will sound something like multiple generations of cassette dubbing. This isn't true in the least. Many sharp transients and attacks, the first thing to go in tape dubbing, will survive multiple generations of MP3 encoding... but the stereo imaging will suffer horrifically, sounds will slide all over the place, and sustained or noisy sounds will become "bubbly". As I believe I saw someone somewhere on this board say in a haiku-like post, "Bubbly Limewire cymbals."

Last edited by kupietz; 2007-12-13 at 01:14 AM.
Reply With Quote Reply with Nested Quotes