Digital Audio Workstation editing software usually has a 'Normalise' function to enable the final two channel mix to be amplified to digital full scale or a nominated lesser level.
I believe that it may be better to not go to digital full scale, but instead go to one or more LSBs less than full scale.
Question is what minus dB level value WRT full scale corresponds to a few LSBs below full scale.
Any further thoughts or experiences in this matter ?.
Eric.
I believe that it may be better to not go to digital full scale, but instead go to one or more LSBs less than full scale.
Question is what minus dB level value WRT full scale corresponds to a few LSBs below full scale.
Any further thoughts or experiences in this matter ?.
Eric.
mrfeedback said:
Every bit 'spare' corresponds to a factor of 2 (amplitude) extra headroom. This is equivalent to 20 log 2 = 6.02dB.
Cheers
IH
Eric, why are you wondering about this? All digital equipment can handle 100% Fs and when you do this normalization you won't get 101%, no risk of clipping.
Ian, eric means how many dB's is 32765 instead of 32768 = 0.000795 dB -> not much
Ian, eric means how many dB's is 32765 instead of 32768 = 0.000795 dB -> not much
Apart from that: Every scaling operation will increase noise.
I don't see much advantage in doing this anyway (apart from equaling out the loudness of several pieces on one CD for instance).
If your peaks are really hitting full scale for several samples per half cycle, no scaling will help you to restore the clipped peaks.
Regards
Charles
I don't see much advantage in doing this anyway (apart from equaling out the loudness of several pieces on one CD for instance).
If your peaks are really hitting full scale for several samples per half cycle, no scaling will help you to restore the clipped peaks.
Regards
Charles
I think the noise increase is not a big problem because the noise level in the music itself is probably higher caused in the recording and mixing process.
- Status
- Not open for further replies.