GigaStudio Dithering and RME Hammerfall DSP Digiface

[Mega]

Hi...I just upgraded my sound card with the RME Hammerfall DSP Digiface, and I am very pleased with its ease of use, low latency, and excellent sound quality.

I did have a question about the GigaStudio dithering setting for this soundcard. The GigaStudio help says that dithering to 16 bits should only be used when the card doesn\'t support 24 bits. Since the RME Digiface card supports 24 bits I decided to disable the dithering I had used before my upgrade.

But then the GigaStudio help confusingly goes on to say that if you are using a separate D/A converter, then dithering should be enabled to avoid pops. And I am using a separate D/A converter, a Frontier Design Tango24 hooked up to the all digital RME Digiface via ADAT. So should I enable dithering or not? For now I have left it disabled and have not heard any pops.

In fact, this also leads to some questions about dithering in general. I thought GigaStudio did not support 24 bits, only 16 bits. But I guess I have come to realize that this just means 16 bit samples, right? The mixing and effects might be done at higher bit levels for better sound quality, right? So dithering should probably only be used if the soundcard itself only supports a maximum of 16 bits. And then GigaStudio has dithering type 0 and type 1. What does that mean and when should a particular type be used?

Thanks for any info!

[JonFairhurst]

The simple answer is \"don\'t use dithering in your case\".

You are right about Giga using 16-bit samples and creating a 24-bit mix. The dithering should only be used in the case of a 16-bit D/A.

There are three basic ways to get to 16-bits: truncate (just hack off the lower bits), round (add 0.5, then hack) and dither (add noise, then hack). The dither method reduces \"banding\" by randomizing the point at which the bit cutoff occurs. In video signals the banding due to truncating/rounding is really evident. With a slow ramp you literally see bands of different brighness across the screen. Dithering helps mask this effect.

The two dither types would likely be two different noise profiles. I guess it comes down to which one your ears prefer when using 16-bit audio. I\'ve never seen anything published by Nemesys/Tascam that defines the profiles. (White noise / pink noise / Giga noise?)

[Nick Batzdorf]

The principle is simple: add dither whenever you move to a lower bit rate, as in 24-bits to 16. It doesn\'t sound like you\'re doing that, so you turn it off.

One thing to be aware of, as Jon says, is that even though the files are 16-bit, the mixer will generate more bits as soon as you move a fader.