Garbage In, Garbage Out?
 

Hey all.

I've been broadcasting for nearly 2 years now at 128 kbps. Over that time, I have been converting most of my mp3s to 192 bit. When playing the songs through my station, and listening to my stream, it seems as though the 192 bit mp3s have a much better quality about them, despite the fact that I'm still only broadcasting at 128.

Is this improvement in sound quality for real, or is just in my mind, knowing that the source song is improved quality? Does it come down to the proverbial "garbage in, garbage out"?

By all respects if you take a 320k (standard CD quality) and reduce it to 128k you already take away some of the sound quality. Then, when it's re-encoded yet again to the server you start off with less quality and thus reduce it even more.

My thought is you're right.

Converting to 192 bit from what source (re-ripping)? Just curious because you can't replace the missing data by simply converting from 128 to 192...

Most songs were originally ripped from CD at 128. As I strived to improve sound quality, I've converted most of them to 192, using Audacity to convert them to WAV, then a generic mp3 converter to put them back at mp3 format at 192. In all cases, they sound much better listening to the source file after conversion (I was also able to manipulate volume, bass, etc), and thus, they seem to sound better when listening to them play over the stream.

Edited to Add: Some songs needed no volume/bass/treble manipulation, and they still sounded better through streaming, after conversion.

Even converting MP3 - WAV - MP3 will not replace the data that was removed during the original encoding process. I will give this Audacity a try and then run the tracks through our analyzer to see if you have in fact stumbled upon the goose that lays the golden eggs.

Thanks Ozzie. I appreciate the help!

Audacity can be downloaded at this site.

re-encoding
 

Okay here is a good way to test the MP3-WAV-MP3 deal (re-encoding to a higher bitrate).

Pick an MP3 file you like that has some good high end.

Convert to WAV and re-encode at 192Kbps. Or you can just re-encode to 192Kbps directly (the WAV conversion is a wasted step).

Open both files and notice that the 192Kbps file has lost some amplitude. Look at both files in the spectrum view. They should look almost identical. The data that the original compression removed has not been replaced.

To prove the data has not been replaced do a re-encode to 32Kbps and then back to 192Kbps. Look at the 32 and 192 files in the spectrum view. Any noticable changes in the spectrum view are "normally" due to artifacting. The data is still missing and you end up with a much larger file (wasted space).

To obtain a "true" higher bitrate you need to re-rip and re-encode at the higher rate. Right now you are just wasting HDD space.

My 128Kbps file is 5,403K
the 192Kbps re-encode is 8,105K

Thats 2,702K of wasted space!


This forum has a little talk about it.

http://www.cdrom-guide.com/forums/sh...d.php?t=103131


All testing was done using Audacity. Thanks for the link BTW!

When ripped from any outside source cd, deck, etc.and converted to mp3,I usually convert to 320kbps,before I do this I usually run the treble filter over the file, as some treble is lost on conversion(according to syntrillium, if memory serves me right) to mp3,this makes for a sharper tone.

I find that 320kbps with the treble filter pass sounds every bit as good as the wav format.
the 192kbps is also acceptable.

Also note some cds and vinyl are recorded at different levels,some modern cds when ripped normally are nearly always at skip level especially dance -club cds(distortion)this is the quirk of the mixer,some who think that dance should be so loud as to skip.

Did some images of the re-encodes (used GoldWave this time).

The _2 files are snips from the _1 files.

Also note on the 128 to 192 re-encode there is additional data missing (as one would expect from yet another round of compression).

http://www.wrzwaldo.org/Audio/

An interesting page on Perceptual Coding.

Ozzie,

Thanks a bunch for that info. While the spectrum may very well show that additional conversions result in additional data loss, to my ears, most of the mp3s that I've re-encoded to 192 sound much better. That might not make much sense, given the info that you've given me. But then, perhaps it's all a personal preference. After all, we all hear things differently. Most of us differ in our sensitivity to certain frequencies. Maybe that's why I hear 192 as better quality. They just seem so much clearer and cleaner. And going back to my original post "is it just in my head?", well I think you just proved that. LOL

Anyway, thank you for the link to that discussion forum. Very interesting info I must say. And again, many thanks for your look into this question. It's nice to know that there are people with the "nuts and bolts" knowhow of this stuff to help others.

Oh by the way, I'm well aware of the extra step that MP3-WAV-MP3 entails. Apparently, I needed some extra file to go with Audacity in order to do a direct MP3-MP3 conversion. I guess I was just always too lazy to ever go get it and install it. LOL

You just need the LAME encoder dll (free also) put in the audacity folder!

http://lame.sourceforge.net/

Also please note that I was in no way trying to knock the fact that you were hearing an improvement. It is possible you were. Just wanted to shed some light on audio compression and the possibility that you may be wasting disk space.

Oh I understand that Ozzie... but HD Space is nowhere near a problem here.

By the way, in your honest opinion, what did you think of the Audacity program? I think it's pretty darn sweet for being freeware, yanno?

I was rather impressed by it. And open source to boot! I am a big fan of GoldWave but Audacity is a good product also.