masterhse wrote on Tue, 06 November 2007 10:34 |
So my first two questions:
1. How can you tell if a track has been truncated or dithered other than by listening to artifacts? 2. What are the advantages of dithering before applying noise reduction?
|
1.
Use a bit-meter, and look for stuck bits. Bit-Police is a good one, it actually shows the percentage distribution of each bit number. If there is an even distribution especially around the LSBs, then there was most likely dither added. Izotope Ozone has a bit-meter as well. Wavelab also has one built in. Bit-Police is free and VST.
Also an FFT display with a very low noise floor setting and at least 16000 bands sometimes helps confirm suspicions if there was dither or not, depending on what kind of dither it was. You'll see the dither in the display as the music is fading in or out, it is a constant , even, broad-band noise, unless very extreme noise shaping was used. In that case, you would look for a shape of noise that is generally constant underneath all of the actual signal activity.
2.
Dither provides a more or less even distribution of the noise that is there in the first place. This makes it less likely to have artifacts when you are further going to reduce the noise floor with dsp.
Dither helps to mask the noise that is going to be removed, it sort of congeals with the noise and then soaks it up, allowing for smoother low levels of sound.
By dispersing the noise first with dither, it doesn't require such extreme settings from the Noise Reduction software.
And yet another way to describe it is that dither shakes loose the quantization noise that is latched on to the signal you want to keep, so by shaking the distortion loose, it is then easier to come by and sweep it up afterwards along with the other noise that you are trying to clean up.
Oh, yeah, and in case you haven't figured it out,
I'm in.