HD 2 -> how many VOICES ? And how does PT "count" them ?

[LukeFromBerlin]

Hi

I have a HD 1 running on a pci-e core accel card (MAC) and 2 Interfaces and am considering to buy a second accel card. I guess what I have then would be called a "HD 2" (?)

Now I am trying to figure out how many VOICES (not channels) I would have then. If I check out Avid's comparison charts, they tell me:

"Protools HD" : Sample Rate 44/82/192 = 192 / 96 / 36 voices.

But which HD stage is this referring to ? A HD 3 ?

Now in the other line it says: "HD 1": Sample Rate 44/82/192 = 96 / 48 / 18 voices

OK, that's my current setup. But what about HD 2 ?



Does my number of voices really double with the second card ? And nothing happens anymore when buying a third Accel card ? Seems unlikely.

Hope someone can help as the number of voices thing is important to me (often working with 88_2 KHZ and a bunch of RTAS Plug ins)

AND: perhaps an expert can finally tell me HOW Protools really adds or "uses" it's voices ?

ONE voice per output channel no matter where it is routed to ?
ONE voice per created or only per used (routed) AUX channel ?
ONE more voice per channel that is input monitoring at the same time ?
ONE voice for EACH RTAS Plug in ?

Would appreciate it if someone knows
(sorry for being lazy and not testing back & forth an always re-checking at what point PT says "no more voices available". And I wouldn't know any better way of finding out what exactly uses up voices...)

THANX

[cneal92]

The number after HD refers to the # of cards you have. An HD1 has a single core card (accel or not). An HD2 has a core card and an accel (or process) card. The interfaces don't matter, except in regards to how much simultaneous I/O you have, and you can only connect two interfaces per card, so with an HD1 you can't have more than 32 I/O.

So you currently have an HD 1 (96/48/18), and if you add an accel card you can attain the highest track counts of (192/96/36). So at 88.2, you would get 48 voices on the HD1 and 96 on the HD2. Yes the number of voices goes up with the addition of a second card and does not increase with additional cards.

One playable track = one voice. If an audio track has either an output to a physical output or a bus, or a send to a physical output or a bus it would use one voice. Routing that track to multiple physical outs or buses does not use more voices.

There are some situations that would cause extra voice usage, regarding using RTAS plug-ins on audio tracks after a TDM plug-in and using RTAS plug-ins on Aux Input & Master fader tracks. These cause additional use of one voice (per channel) to get the audio off the card and to the CPU and an additional voice to get the audio from the CPU back on to the card. Also using the side-chain on RTAS plug-ins can cause extra voice usage. There are other situations that could arise as well.

Rule of thumb, it is fine to use RTAS on Audio tracks, as long as they precede TDM plug-ins, if any. On Aux In and Master Faders only use TDM plug-ins unless absolutely necessary. The use of RTAS in these situations causes increased latency and extra voice usage. So I still will use some of my RTAS only plug-ins, like Waves IR-1 on Aux In tracks as a Reverb return, but I know it will take extra voices to do so. The latency induced is taken care of with ADC. On HD systems too, your HW Buffer affects this latency on Aux In tracks, so if your ADC is not able to account for the latency, make sure your HW Buffer is not set higher than needed.

You can have upto 256 tracks in your session and use explicit voice allocation to get more than 192 tracks that "can" play, although only 192 can be playing at once (at 44.1/48 kHz of course), so if you had a Tamborine and a Shaker that didn't play at the same time one in the verse the other in the chorus, you could explicitly assign them the same voice and those two tracks would only use one of your available voices.

Sorry, I don't have time for a full detailed explanation of all that.

[Rail Jon Rogut]

... and QuickPunch needs an additional free voice for each record armed track (so 2 per record armed track).

Rail

[LukeFromBerlin]

THANK YOU ! Especially @ CNeal, this is great help and it would have taken me forever to collect all this information from various sources...

I think the answer was very thorrow and I appreciate you taking the time for it

[kooz]

Originally Posted by Rail Jon Rogut

... and QuickPunch needs an additional free voice for each record armed track (so 2 per record armed track).

Rail

My initial reaction to this was "WHAT!?!?!?!?!" but it makes perfect sense.