There has been some discussion on daw-mac email list about missing voices and what had eaten them up. Folk chipped in that using RTAS plug-ins can cost voices. So I thought it would be good to outline the conditions under which this can happen.
Basically in the TDM architecture which is the core on which the HD systems are built, there is one law that cannot be violated: Every signal entering or leaving the TDM Environment requires a Voice to do so; each Voice is one-way only.
So in more detail, if you have a mono audio track, the audio coming from the disk requires a Voice to enter the TDM Environment. Audio tracks are the only track type that will grab a Voice by default upon creation, and keep it as long as it is active. Make a wider track (stereo, LCR, etc), and an additional Voice is grabbed per channel added. Therefore, a 5.1 audio track will use 6 voices upon creation. On the other hand Aux In, Instrument, and Master Fader tracks do not utilise a Voice by default.
When it comes to plug-ins, TDM plug-ins do not use a Voice ever, because TDM Plug-ins exist solely within the TDM Environment and so you don’t need to leave the TDM environment to use a TDM plug-in. Slightly more surprising is that hardware attached to the HD cards like Digidesign interfaces is considered part of the TDM Environment, since they are directly connected. This means using a Hardware Insert would not grab a Voice, for entering or leaving.
Moving on to RTAS plug-ins, they are powered solely by the native host. In order for the native host to do this, the signal to be treated must be sent away to the host CPU. To do this, the signal must either originate from the native host, and the only way to do this is a hard disk feeding an Audio track, OR, it must leave the TDM Environment, get processed in the host environment, and then return to the TDM Environment. It does this through those one-way gates, known as Voices.
So when you have a signal in the TDM Environment, and you want send it out for RTAS processing, it must use a Voice to leave, and then another Voice to re-enter, which means to use an RTAS plug-in on a mono track costs you 2 voices. Multi-channel tracks will cost you more voices.
However as with most rules there are exceptions and this rule is no different. If the track is an Audio track, and is being fed from the hard disk (NOT via live input), and you have an RTAS plug-in as the first plug-in the signal goes through, the audio can simply pass from the Hard Drive to the CPU, and then onto the TDM Enviornment, so only costing one voice. You can have a number of RTAS plug-ins in series as it it is only at the point at which the audio goes into the TDM environment that it will cost you a voice.
So there are some tricks to employ to keep the voice count down.
- Try using RTAS on Audio tracks that are playing off Hard Disk.
- Try to use TDM plug-ins on Aux tracks.
With the ever increasing voice counts on TDM systems we have become complacent but with the growing use of RTAS plug-ins and higher channel counts when working in Surround especially when you have an HD Core system, voices can be an issue you need to plan for. This is now a serious consideration in deciding whether to use an HD or LE system.
Following on from the thread on daw-mac, Jerome Boiteau, who raised the issue, has just posted…
I just installed two 8 core LE systems. They totally beat our HD3s. In every way. We are mixing a TV series on one of the 8 core. A 5.1 mix AND a stereo downmix in the same session. Multiple 5.1 and stereo busses, C4, RenEQ and DeEssers on each of the 12 dialog tracks. Three UM226 (stereo to 5.1 upmix from waves), Rencomp+ 2C4 + L1 on the 5.1 mix bus, and the same on the stereo mix bus, and HD video playback. 5.1 and Stereo Dorrough meters. The system usage never goes above 29%
Jerome Boiteau
Studio Expression, Quebec, Canada
Very interesting developments. Have we got to the point where it will be better to dub on host based systems using a large multi-processor machine running Pro Tools LE with either DV Toolkit 2 or even better The Complete Production Toolkit and drop the HD hardware and its significant cost to boot.
Please add your thoughts and experiences to the comments below.
This is the first in a series in which I will outline the basics of how to use Digidesign’s Pro Tools systems to work to picture. First of all, we look at the decisions you need to make in setting up your system. There are three basic ways to work to picture in Pro Tools.
