Hi

This is my first post, Great community by the way, and here is my question:

In the spectrum graph(Made with Audition), there are 3 files (Only the Left Channel is shown):
1.-wav file with 44100hz sample rate.
2.-mp3 encoded with lame (--vbr-old -q 2 -b 224 --preset extreme --resample 48000).
3.-mp3 encoded with lame (--vbr-old -q 2 -b 224 --preset extreme).

In the spectrum graph, the resampled mp3 looks more saturated than the other not resampled mp3 but not as saturated as the wav file. Does this mean that the resampled mp3 has a fuller,richer or better sound, or is just my imagination?, I have a mp3 player that supports 48000hz files, Do you recommend me to resample?, or Its just a waste as I read in other posts?

Here is the spectrum graph:

Spectrum Graph

Thanks in advance for all the replies.

Jose.

Jose, from my experience with encoding mp3's with LAME you will create an mp3 that is always a couple decibel louder than than your source file. I don't actually know why this happens in LAME, but I've encoded CD's into LAME just to find out that the song was mysteriously amplified by +1 or +2dB. I also would not recommend resampling. If you have a file that is originally at 44.1kHz its best to keep it at its original rate. Upsampling to 48kHz from 44.1 doesn't actually increase the quality of the song but can in fact reduce it somewhat. Resampling to a higher number cannot make a song richer because it doesn't add any new sound into the song. i.e. a 12kHz tone in the source will still be a 12kHz tone in the upsampled version...so nothing would actually change sonically.

-808

Looking at the spectrum, or any other kind of graphical analysis, is not a useful way to compare any lossy encoding with the original. Regardless of the differences you see, they have little or no relationship to the perceptual qualities - what you hear. mp3 and other lossy encoding are perceptual encodings, based on studies of human hearing, not data reduction per se.

Ok, Thanks to BGonz808 and AndyH-ha for the replies, indeed it was my imagination.

Jose.

The spectrograms have different ranges. One goes up to 22kHz, the other to 24kHz. That's why they look different. Both encodes have about the same bandwidth, topping out at around 22kHz.

I agree with you that the encoded ones cut at aprox. 19khz, I was saying that the one that is resampled to 48000(2) looks more saturated than the one that it's not resampled(3)