Skip to main content

Topic: Size based test? (Read 4430 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • Birch
  • [*]
Size based test?
Has anyone done a listening test comparing codecs where there is a target file sizes instead of a target bitrate? 

I'm interested in good sound, but that is tempered by the fact that I only have 512MB on my player.

Maybe it's a bad idea.  If so, let me know why.

Thanks,
Birch

Size based test?
Reply #1
Well, I think it's the same, because size is just a consequence of bitrate.

size = (bitrate / 8)  * length, in mp3 at least. I think WMA is different.
"You have the right to remain silent. Anything you say will be misquoted, then used against you."

  • Omion
  • [*][*][*][*]
  • Developer
Size based test?
Reply #2
Quote
Well, I think it's the same, because size is just a consequence of bitrate.

size = (bitrate / 8)  * length, in mp3 at least. I think WMA is different.
[a href="index.php?act=findpost&pid=230205"][{POST_SNAPBACK}][/a]

I don't see how WMA could be different. Bitrate is "bits per second". There's no way around it. If you have x bits, in y seconds, then you have a bitrate of x/y bits per second.

The only oddity I can think of is that sometimes (or most times, or something) the container overhead is not used in the bitrate. So the file may be a tad bit bigger than the result of the above calculation. Container overhead tends to be really small, so it's nothing to worry about.

So yes, every listening test of codecs at the same bitrate are also at the same file size (for a given file). I assume you don't mean the same file size no matter the time duration. That would be a bit odd...
"We demand rigidly defined areas of doubt and uncertainty!" - Vroomfondel, H2G2

  • Gabriel
  • [*][*][*][*][*]
  • Developer
Size based test?
Reply #3
The only difference with wma is that Microsoft is considering that 1kbps == 1024bps, whereas everyone else is considering that 1kbps == 1000bps

  • kjoonlee
  • [*][*][*][*][*]
Size based test?
Reply #4
Quote
The only difference with wma is that Microsoft is considering that 1kbps == 1024bps, whereas everyone else is considering that 1kbps == 1000bps
[a href="index.php?act=findpost&pid=230226"][{POST_SNAPBACK}][/a]

If you do this, then you can lie that your 128kbps files are 125kbps files.

  • Birch
  • [*]
Size based test?
Reply #5
Ok, my bad.  Ha, it makes sense - *bit* *rate* LOL 

  • Gabriel
  • [*][*][*][*][*]
  • Developer
Size based test?
Reply #6
Quote
If you do this, then you can lie that your 128kbps files are 125kbps files.

Why do you think wma "128kbps" files are in fact 131kbps?

  • amarillo
  • [*]
Size based test?
Reply #7
so if file size follows that rule then if two formats are in the same bitrate would use the same space?
if thats so then what is the real use of compresion?(it sounds kinda dumb but it eludes me) essentiela wouldnt all formats be the same?(dont hate me im just asking cause i dont usually make listening tests)

  • Dibrom
  • [*][*][*][*][*]
  • Administrator
Size based test?
Reply #8
Quote
so if file size follows that rule then if two formats are in the same bitrate would use the same space?
if thats so then what is the real use of compresion?(it sounds kinda dumb but it eludes me) essentiela wouldnt all formats be the same?(dont hate me im just asking cause i dont usually make listening tests)
[a href="index.php?act=findpost&pid=230744"][{POST_SNAPBACK}][/a]


If the compression were lossless ( like ZIP or FLAC or something ), then yes.  But the kind of compression you're talking about is lossy.  This means that you will not get the same value from the output that you put into the input.

For something like knowledge based content, this would be very bad, but for media compression where small degredations in visual or audio quality are allowable, it's a useful thing.  However, most people would prefer to preserve as much quality as possible at a given compression ratio though, which is where quality differences in codecs come into play.  Some are better at which parts of the original they throw out than others.

Edit:

This is why Roberto's tests are designed to test the quality of a codec at a given bitrate.  For lossy codecs, where quality is dynamic, this is the kind of thing you want to test.

Size based comparisons only make sense if the quality is static, like if you're using a lossless codec such as FLAC or APE, etc.  There are already numerous comparisons online for this though, and you can run these types of tests automatically.
  • Last Edit: 01 August, 2004, 12:43:33 AM by Dibrom

  • kwanbis
  • [*][*][*][*][*]
  • Developer (Donating)
Size based test?
Reply #9
Quote
The only difference with wma is that Microsoft is considering that 1kbps == 1024bps, whereas everyone else is considering that 1kbps == 1000bps
[a href="index.php?act=findpost&pid=230226"][{POST_SNAPBACK}][/a]

and why is this? isn't this based on the bynary sistem? like in megabyte = 1024 bytes, and not 1000 bytes.

  • sven_Bent
  • [*][*][*][*][*]
Size based test?
Reply #10
Quote
so if file size follows that rule then if two formats are in the same bitrate would use the same space?
if thats so then what is the real use of compresion?(it sounds kinda dumb but it eludes me) essentiela wouldnt all formats be the same?(dont hate me im just asking cause i dont usually make listening tests)
[a href="index.php?act=findpost&pid=230744"][{POST_SNAPBACK}][/a]


Q U A L  I T Y

Y kbits of endxoder X might sound better then Z.

that why you are doing listening tests

remmber that we arenot talkign about bit perfect compression so chocie has to be made about discarding information. some encoder do this better then others
  • Last Edit: 01 August, 2004, 02:26:34 AM by sven_Bent
Sven Bent - Denmark