MP3: VBR may be better than CBR

[Bigtop]

Okay, apparently I may have not been in AHS very much yesterday, primarily because I was too busy with the stuff on the 1,100 miles of "premium" auto racing in the United States, along with the big mile-and-a-half Japanese Derby (the Tokyo Yushun), as it appears that on May 30th it was big day of racing - the most-watched Japanese horse race of the year, (one of) the largest sporting events of the year in the world overall (the Indianapolis 500), and NASCAR's longest race in the calendar (the Coca-Cola 600 at Charlotte).

But what I'm talking here isn't all about the racing - in fact, this is much more to the audio talk that we are usually familiar about: MP3s. And over the weekend, I've been bringing up minds about possible changes into my MP3 fleet. And the right time to do something like that is on this Memorial Day weekend, where I've got three days to relax and 1,102.5 miles of "premium" racing excitement - the three races that I talked about, plus the Metropolitan Handicap (the $500,000 "Met Mile" out of Belmont Park east of New York City).

While I was watching those 1,100 miles of auto racing on Sunday, I've decided to actually start replacing the Ayumi Hamasaki fleet from the 320kbps we've been familiar about to something that might actually be even better: variable bit rate!

Apparently, as I researched along with that I've got a really fast computer, as well as on an upload of JPopsuki that I actually uploaded an "inefficient" version of a converted Arashi's "All the BEST! 1999-2009" from FLAC to MP3, only to find out that it was not the right method! I found out that I was using "joint stereo", which is not efficient compared to pure stereo, and I decided that I deleted the torrent to declare it is void all because of an improper encoding method. (And I did the same for Sayuri Sugawara's "Sunao ni Narenakute", as it was encoded improperly.)

So then, I decided that I checked the profile of the user who reported that it was not a pure MP3, and I found out that there was a link for something that is actually changing me how I'm going go enjoy my MP3s anywhere: it was the popular LAME MP3 encoding system and how I can actually enjoy MP3s without spending a lot into my iPod storage!

That was the "variable bit rate" system - it could actually get me very-close-to-320kbps-quality while saving space on my storage!

Often here at AHS, people upload 320kbps of audio, but what they might not be realizing that it's actually something called "constant bit rate", or something called "CBR", which actually have quality rates that are rarely better than the best VBR setting, such as V0 rating. And apparently, going 320kbps might be inefficient compared to V0! Take a look at "synthetic charts" (click on image for more information):


And even JPopsuki recommends V0 VBR over 320kbps CBR:

MP3 in V0 (LAME VBR setting with an average bitrate of 250kbps) is our preferred lossy format. It gives the best quality at a resonable filesize.

I think now I should recommend V0 VBR over CBR (e.g. 320kbps) - each setting myg have its own compromise, but V0 VBR might be the best possible:

• CBR at 192kbps might have good file sizes, BUT quality might be degraded compared to 320kbps CBR or VBR.
• 320kbps CBR is currently the "common" standard for uploading music on AHS (whether it's Ayu or not) as well as JPopsuki and other sites, but along with 192kbps CBR and other CBR settings, bit rates and sizes are quite constant in (ratio) relationship. As such, the higher the bit rates go, the larger the file size is, and often file sizes are quite large, and it can really be a hassle when space matters. (For example, Ayu's RnRC - which is Rock'n'Roll Circus - our 320kbps rip was 150 MB!)
• V0 VBR actually gives the highest efficiency in relationship between quality and file size. How it actually works is that the encoding software (such as LAME) scans for output quality, then goes on how many bits each segment use, and as music have variable dynamics, VBR has its own dynamics: it sets up how many bits it put on each segment. This provides possibly the highest quality with the least possible bitrate and file size possible for its quality. As such, in just about any case, V0 VBR has a better quality-file size ratio compared to 320kbps CBR.

Overall, I should recommend V0 VBR over 320kbps CBR. V0 VBR offers a better quality-file size ratio over 320kbps CBR, and many people won't even recognize that the quality is (nearly) the same between V0 VBR and 320kbps CBR.

I particularly recommend the LAME encoding system over other encoding systems because of it's open-source (LGPL), highly-advanced technologies, such as the the ability to test actual output quality to ensure the desired quality level is always achieved in VBR, making it the best VBR model implementation compared to other encoders which do VBR encoding based on predictions of output quality, and making LAME the best in its class compared to its rivals (particularly in the mid-high quality bitrates).

You know what? After all that research, I've even decided to convert operations of Ayumi Hamasaki's albums from 320kbps CBR all the way to V0 VBR! I decided to convert FLAC files of all of Ayu's albums - A Song for XX all the way down to Rock'n'Roll Circus, plus A BEST, A BEST 2 and A COMPLETE, but not those remix albums, plus a few singles since 2007, and the future releases that I am looking forward to - MOON/blossom (48th), the 49th and 50th singles are going to be VBR in my iPod soon!

What do you think about VBR over CBR? And do you actually prefer VBR over CBR? Tell me (and us) what you think.

Some links:

• Hydrogenaudio Knowledgebase - LAME, VBR, CBR
• LAME - http://lame.sourceforge.net/
• Audacity (the software I used to do converting, in tandem with LAME) - http://audacity.sourceforge.net/