BeSweet Downsampling

[JReiginsei]

I transcoded from vob to abr mp3 128 twice using latest besweet beta. One time at 48kHz and the other time at 44.lkHz. The filesizes only differed by a couple hundred kilobytes. I thought by downsampling you were supposed to save a lot of space? The 48kHz is 103,898KB the 44.1kHz is 103,615KB. Here is the log for the 44.1kHz, I used the same settings for the 48kHz as well - SSRC of course. Did I do something wrong or by downsampling you only save a few KB's?

BeSweet v1.5b4 by DSPguru.
--------------------------
Using azid.dll v1.8 (b825) by Midas (midas@egon.gyaloglo.hu).
Using hip.dll v1.13 by Myers Carpenter <myers@users.sf.net>
Using VOBInput.dll v1.2 by DVD2SVCD (http://www.notrace.dk)
Using Shibatch.dll v0.2 by Naoki Shibata & DSPguru (shibatch.sourceforge.net).
Using lame_enc.dll v1.28 (4/6/2002), Engine 3.92 <http://www.mp3dev.org/>.

Logging start : 11/05/02 , 18:34:25.

c:\DVDTools\BeSweetv1.4\BeSweet.exe -core( -input c:\dvd\stream.lst -output c:\divx\temp.mp3 -substream 0x80 -logfile c:\DVDTools\BeSweetGUI0.6b61\BeSweet.log ) -azid( -n1 -s stereo -c normal -g 10db -L -3db ) -ota( -G max ) -ssrc( --rate 44100 --twopass ) -lame( --scale 1 --alt-preset 128 )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : c:\dvd\stream.lst
[00:00:00:000] | Output: c:\divx\temp.mp3
[00:00:00:000] | Substream ID: 0x80
[00:00:00:000] | Floating-Point Process: Yes
[00:00:00:000] | A/V Delay found : 0msec
[00:00:00:000] +-------- AZID -------
[00:00:00:000] | Output Stereo mode: Stereo
[00:00:00:000] | Total Gain: 10.000dB, Compression: Normal
[00:00:00:000] | LFE levels: To LR -3.0dB, To LFE 0.0dB
[00:00:00:000] | Center mix level: BSI
[00:00:00:000] | Surround mix level: BSI
[00:00:00:000] | Dialog normalization: Yes
[00:00:00:000] | Rear channels filtering: No
[00:00:00:000] +------ Shibatch -----
[00:00:00:000] | Source Sample-Rate: 48.0KHz
[00:00:00:000] | Dest. Sample-Rate: 44.1KHz
[00:00:00:000] | Attenuation : 0.0db
[00:00:00:000] +-------- LAME -------
[00:00:00:000] | 'abr 128' preset is used
[00:00:00:000] +---------------------
[01:47:42:816] Gain of 6.0dB had been asserted to file.
[01:47:42:816] Conversion Completed !
[01:47:42:816] Actual Avg. Bitrate : 131kbps
[00:41:55:000] <-- Transcoding Duration
Logging ends : 11/05/02 , 19:16:20.

[JohnMK]

You don't save file size at all. It's bitrate over time. It doesn't matter how accurately or how many samples you take over time. It's the bitrate. And you specified concretely that it should be 128kbit per second, so don't be surprised if that's what you get. :D

[JReiginsei]

Hmmm, thanks for your reply. You just reconfirmed something that I knew all along...I'm a dumbass. I think I'll still downsample to 44.1kHz though, because I read that lame 3.92 is optimized for 44.1kHz.

[scmccarthy]

Hi, I'm pretty new.

Do you mind if I remind you what the advantage of downsampling is?

The advantage is similiar to using a lowpass filter. Even though 128 bits per second determines the file size, the actual amount of compression is reduced. Effectively you are starting from a smaller file so lame does not have to compress the file as much to get the target bitrate. When you use a lowpass filter high frequency detail is thrown out with a better encode of mid and low frequency audio for every bitrate. That is true for downsampling too.

This is probubly recommended for 128 kbps because at that bitrate high frequency information is not faithfully encoded anyway. Also, it is a good idea to match the frequencies that get encoded with the frequencies your speaker can reproduce. Meaning that using a highpass filter to eliminate very low frequencies might be a good idea too.

[DSPguru]

filtering the signal (lowpass, highpass) has indeed effect on the total quality. but in fact, the biggest factor derives from the fact that 44.1khz & 48khz uses completely different filter-banks and each mode needs its own tuning for reaching HQ encoding.