Sample rates and latency in HD

[Irene]

I've gleaned from Greg and Nika that if the Digi interfaces are really top notch the sample rates for getting audio into PT should not make any difference.

As running sessions at 96kHz is best
for the plug-in quality, it makes sense to record at 96kHz then, even though the audio sampled should sound the same at 48kHz - it saves upsampling.

As far as the duc goes most people seem to reporting very noticeable differences at 96kHz. So with each glowing report it actually reflects badly on the Digi interfaces...
The upside is that if there are not so good the latency should be better and live recording will be easier...anyone know what the latency times are with the new interfaces?

thanks

Renie

 

[Greg Malcangi]

Hi Renie,

Only some plugs will sound better running @ 96kS/s. Namely, those ones that create frequencies above about 20kHz and have to apply filtering to stay within the Nyquist point. EQ plugs are an obvious example of this, although you may not hear much of a difference with say the Sony EQ as it's filtering is the best I've heard.

I still haven't heard the 192I/O but from general impressions it seems that Digi have attempted to keep the cost down relative to it's features. An area that appears to have been sacrificed is the quality of the filters at lower sample frequencies. This would explain why people are noticing such a difference using the different sample rates.

As far as latency is concerned I don't know the answer. In theory the latency should be a bit lower at 96kS/s but then on the other hand PT has got to move around audio files which are twice the size, this obviously isn't going to make things faster. My opinion is that Digi has taken latency seriously as it's one of the main selling points of PT over host based (native) systems. In a real world situation though, only time with the new units is going to tell us if there is much of a difference.

Greg

 

[Irene]

Greg,

Thanks, that's very interesting info.

So if I use Sony EQ for EQ exclusively I may not need to record higher than 48kHz ? I will be using my Trak2 for A/D so I would hope that I don't need to go to higher sample rates to get quality filtering...

You mention EQ primarily but are there other types of plug-in specifically that would benefit from higher sample rate sessions?

BTW Are you going to buy a new Mac for this upgrade - are you planning on using SCSI or Firewire?

thanks

Renie

 

[Greg Malcangi]

Hi Renie,

Basically any processing of your audio that might give results beyond the Nyquist point has to apply anti-alias filters to prevent this from happening. This is where the audio quality problems arise. If a plug in or ADC doesn't have the very best filters @ 48kS/s there should be a noticable improvement @ 96kS/s. The better the filters @ 48kS/s the less of an improvement will be heard @ 96kS/s, right up to the point of having theoretically perfect filters where there should be absolutely no audible difference between 48kS/s and 96kS/s. However, the perfect realtime filters @ 48kS/s does not exist. Your Trak2 is pretty good but only a real world A/B test would tell if that's going to be as good @ 48kS/s as running a 96kS/s session using a lesser converter.

I haven't heard the Sony plug @ 96kS/s so I can't say for sure. But in theory as it appears to have good anti-alias filtering @ 48kS/s there should be less of a difference @ 96kS/s. However, this is only the theory and can't take into account any algorithm changes that Sony make to the plug when they convert it to HD compatibility.

I will be buying a new Mac when I upgrade to HD but I've still got to find a free week or so between now and the end of June in which to make the change. I will probably use both firewire and SCSI drives. Firewire for video and SCSI for audio.

Greg

 

[stedel]

G'day folks. Hi Renie. Heard from Catmixer lately?

The Creamware Pulsar system (their upper-mid range line)has been 96k for maybe two years now and produces AD/DA converters for this format. Cubase VST32 - which interfaces seamlessly with the Creamware stuff, has also been able to handle 96k for at least a year now. The Creamware stuff is like Pro Tools in that it comes on a card equiped with its own processors, in this case they use a number of Sharc chips. Unlike Pro Tools however they use a 32bit floating point system.

My point here is that running these at 96k on a G4 at 540MHz with 240megs of ram has been no problem
-at least up to 24 tracks - and for the normal
type length of a rock pop toon (maybe 5mins max).
Beyond that you're likely to push it particularly if you use a lot of plug ins - but mainly the issue is to do with the amount of RAM you're running. Like Pro Tools Creamware also have a modular system whereby you can buy another card to boost the processing power available.

The Tools HD system has increased the number of chips on its cards, and though they're not Sharc chips (which are also used by Fairlight), they should (you'd think) give you enough grunt to shift the larger files etc around-which BTW are not as big as you might think, although there is a quite noticeable increase in file size.

So advice no.1. Get yourself as much RAM as you can.

Again for me the proof has to be in the quality of Digi's converters. A lot of people out there are using 32bit floating point systems, from Apple, Nuendo/Cubase to Fairlight systems - a format that Digi argue against. (BTW don't be too quick to slam native systems like Cubase, they've come a long way in the last few years).

I gather that from what people are saying here maybe Digi have compromised on the quality available for the old formats like 48k. Is this the emerging opinion?