48khz -> 44.1khz

downsampling required, which of the following is preferred ('less' lossy) ?

1) ssrc_hp (1.30) --twopass --rate 44100 --bits 16 --dither 1 --pdf 1 1

2) fb2k: dsp -> resampler (ssrc - slow mode)

or is there an even better solution in fb2k >0.8.3 ?

must downsample. want optimal quality.

You don't want dither added unless you are reducing the bit depth, otherwise it is just more noise to no advantage.

Are you sure this is not required?
I'm not.
It may work without distortions but there's no guarantee, is there?

Edit: Just a guess: A rectangular dither might be appropriate and safe in this situation.
The noise floor will be 3dB lower compared to full triangular dither.

Edit2:
I just looked at SSRC's source code to check what it actually does concerning dithering and noise shaping.
Unfortunately "noise shaping" is called "dither" by the author.


If however -- like kjoonlee pointed out -- you're interested in converting those to MP3 better use "ssrc --rate 44100 --bits 24". LAME can handle 24bit/sample files.
OR: Just use Foobar2K to transcode. It doesn't really matter what resampler you use. They're both good.
I totally agree to this reply and the following by kjoonlee you got in the other thread.

http://www.hydrogenaudio.org/forums/index....showtopic=48105

The next step will probably be MP3 encoding. I'd skip the 16-bit stage.