Post
by solacerodgers » Wed May 13, 2009 1:25 am
I know this topic has been beaten to death but I just got a chance to read the 3 guides and see what they said about the subject of intersample peaks and dithering in general.
On the topic of dithering its a excellent read and you also get audio examples to hear and see what is going on as well as a excellent guide for using dithering with ozone in almost every DAW but still the only words said about the function of preventing intersampling peaks was this.......
"In the digital world, audio is represented and processed as discrete individual samples or levels. When played back in the real world, these samples are converted to continuous waveforms. In some situations, the nature of this "real world" audio signal could result in clipping "between the samples" even when the limiter is limiting the digital samples. Selecting the "Prevent inter-sample clips" option allows Ozone to intelligently predict the behavior of the analog signal, and prevent, as the control says, any inter-sample clipping."
After much much much reading and research I found out what they mean by "no inter-sample peaks" I only use this in quotes as its more of the DAC's fault then software that causes this but with the MBIT+ dithering that Ozone uses....More on the MBIT+ in a min let's see what happens when you press play on that cd you just put in your player.........
So what commonly happens in the current mass market DAC's is it id designed to do the vast majority of their reconstruction process by employing a digital Finite Impulse Response (FIR) filter design tailored to provide a brick wall low-pass response. This approach overcomes the less perfect slope response issues of pure-analogue filter designs. These FIR filters work as follows, firstly, the audio is up-sampled to a multiple of the original sample rate, then the FIR is used to create the smooth curve in-between the original sample points. Of course different DACs have various ways of doing this and indeed, the finer details of the process are important design elements of the DAC design. So with these flaw's in the design and use Ozone has designed MBIT+ that it is effective at removing quantization error and still perceptually quiet in level. Any differences in the shaping is relatively minor considering we're talking about signals that are going to be heard at -95 dB or more. In short, this is meant to be a positive presentation of two similar "Near Nyquist" shapes and we hope you recognize that as well.
So how does the MBIT+ fix this, by providing a combination of up-sampling and filtering that mimics the operation of an oversampling DAC’s reconstruction process so you get *no quantization errors and a much better reconstruction of the audio from the digital domain to the analog world through the DAC's.
**** IF YOU DONT WANT TO READ THE ENTIRE ARTICLE HERES THE NUTSHELL *****
Yes it does prevent inter-sample peaks when using it correctly even with horrible DAC's, this is still a issuse of truncating the wordlength and a good noise shaping or filtering program when dithering. In short you still maintain the same noise floor with higher detail, not to be confused with increased dynamic range as you can only fit about -96db's into 16 bit audio regardless of bitrate your DAW works in.