Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Where do you get the number 512 samples for Altiverb AU? And with AU plugins, do they cause latency on a record track even if they are used as a send instead of sending the audio through them?

http://www.audioease.com/Pages/Altiverb/HostSpecificDescriptions/Audio_Unit_-_Logic_Audio.pdf

"Altiverb has a built-in latency of 512 samples."

Yes, the plugs cause latency even if they are used as send effects (i.e. on a bus) rather than as inserts. However, latency on reverbs used as send effects isn't a huge problem: it's perceived as pre-delay (i.e. you'll hear the unaffected signal first, and the reverb some time later). The DAW should not be compensating for this delay during tracking (in Logic, you must set PDC to "Audio Tracks and Instruments" and NOT "All" while tracking, otherwise EVERY track will be delayed). During mixing, you MUST compensate for bussed effects (either via automatic delay compensation or by bouncing and manually dragging your tracks around), or you'll have serious phase issues and everything will be out of whack.

When the effect is used an an insert, then you've got to deal with latency during tracking. This would typically be compressors and eqs, which one would conceivably use to tracks a vocalist, for instance. But it could also be a guitar amp sim like Guitar Rig or Amplitube, which can introduce significant latency. This is one of the reasons I've gone to using my PodXT and/or a mic'ed amp.

There are certainly ways around latency on native DAWs, and perhaps 7-10ms of latency doesn't both you (it doesn't both me until I get around 12ms or higher). I know some people who can always hear even a few ms of latency.
 
buyer beware

I noticed that Logic Express was at a reduced price at one of the stores I visited this week-cheaper than the A$499 that I paid for my copy last year.


Just curious where you bought it. I hate to tell you but you did not see
a reduced price on express. What you saw was probably the regular
price of $299. What you paid was a very inflated price. I have never seen
it for over $299 that is why I would like to know where you got it, so I
can make sure never to buy anything from there.


I am also curious what would be obviously improved on Sculpture
and Ultra Beat? I have not noticed anything that obviously needs
improvment.
 
Just curious where you bought it. I hate to tell you but you did not see
a reduced price on express. What you saw was probably the regular
price of $299. What you paid was a very inflated price. I have never seen
it for over $299 that is why I would like to know where you got it, so I
can make sure never to buy anything from there.

Notice the A$499...Australian dollars

http://en.wikipedia.org/wiki/Australian_Dollar
 
. The basic functionality is there and the interface is the easiest to use in the whole industry.

This would be your opinion, I would strongly disagree but that would be my opinion.



If one is serious about audio (compared to someone buying Photoshop because one is serious about graphical design) that person will likely buy a Mac-based Protools HD system. That's because being able to use near-zero latency and TDM/RTAS exclusive plugins.

You make it sound like PTs low latency is the holy grail. It is not since if your
serious about audio like me you have a nice rack of outboard effects and a high quality interface giving you ABSOLUTE ZERO LATENCY. So I just shot down this arugument of PT dominence because of low latency with one sentence and much less of an investment than having to buy all the over rated Digi hardware. I do not use software monitoring for recording and can have any amount of reverb or effect without any impact on latency. If you think $500 for three tape immulation plugins is a good deal, then they really have got you brainwashed.




There are a lot of companies (Eventide, Princeton Digital, Chandler Limited, Soundtoys and McDSP for example) that only develop for the Protools platform. Those are some of the best companies in the whole audio plugin industry! If one wishes to use their products, one *MUST* use Protools.
Well you got one part right, they are SOME of the best. For every plug in that PT has there is an equal or better one for Logic and that is just if your using software. Even Charles Dye admits that there are equivalent plugins for everything in PT (MILAR).

Just because more studios use PT and it is the "standard" really does not
mean that much. I mean more people use windows and PCs so the "More users means it is better" argument does not fly in my opinion. More people drive GM vehicles than BMWs and we know BMW does not worry too much about it.

Logic 8 should be out about the same time as Lepeord, (my prediction). The engineers have so much work to do on this app (it has not had a complete rewrite since 1989) this is why it is taking so long. I am confident that it will be worth the wait. Then it will be the PT users who will be thinking of switching.

Of course this is only "my opinion".


That makes more sense, I would hate to think there are people
that would charge that much. Being an American who does all his
travelling on the internet, I only saw the $ which is does have
some history here in the US.

From Wikipedia:

From 'US'
That $ is a monogram of U and S, which was used as a mark on money bags issued by the United States Mint. The letters U and S superimposed resemble the historical double stroke "$" sign: the bottom of the 'U' disappears into the bottom curve of the 'S', leaving two vertical lines. This double stroke dollar sign has been used to refer to US Currency. Thus, the one stroke design may have been modified to the double stroke design to represent United States currency. This theory was largely popularized by the novel Atlas Shrugged by author Ayn Rand.
 
That makes more sense, I would hate to think there are people
that would charge that much. Being an American who does all his
travelling on the internet, I only saw the $ which is does have
some history here in the US.

From Wikipedia:

From 'US'
That $ is a monogram of U and S, which was used as a mark on money bags issued by the United States Mint. The letters U and S superimposed resemble the historical double stroke "$" sign: the bottom of the 'U' disappears into the bottom curve of the 'S', leaving two vertical lines. This double stroke dollar sign has been used to refer to US Currency. Thus, the one stroke design may have been modified to the double stroke design to represent United States currency. This theory was largely popularized by the novel Atlas Shrugged by author Ayn Rand.

Just to help the etymology along:

http://en.wikipedia.org/wiki/Thaler
 
This is a backwards way of looking at USB vs. FW vs. PCI throughput. (...) In reality (...) The fact that an internal SATA drive can stream faster than a FW400 or 800 drive has nothing to do with the fact that it's internal or somehow closer to the main system bus. An eSATA drive will give you the same throughput. The limitation is the FW400/800 standard, which simply isn't able to offer the throughput. (...) However, this has very little to do with the issue of latency. The problem with USB and FW interfaces is that they all have a safety buffer that keeps things running smoothly. (...) PCI interfaces plug straight into the PCI bus, and therefore don't need the safety buffer, and can generally offer lower latency operation.

You have to really think about what you just said. That's because you just proved my point. Why on earth there is a safety buffer in the first place? The reason is very simple: latency. There has to be a buffer, because the system is too slow to react to a moving target. That's the problem with every system that has moving parts, in this case, a clock, which means the system can only ask for data X times a second.

Think about hard disk, for example. New hard drives have large areal data density, which means that once the proper place has been found, the drive can stream that kilobyte very fast. Faster than ever. But imagine current hard drive at 1800rpm rotational speed. If that kilobyte has just passed the reading head, the drive will have to idle for 33 milliseconds until it can be read. Now tell me: did it take +30 milliseconds to read that kilobyte or was the data instantly available?

That's exactly the same reason why USB "in reality" is not so fast as FW, which is "in reality" not as fast as PCI. You can give instructions to USB device only once in a millisecond. FW device can be told what to do 8 times a millisecond. PCI device can get instructions a thousand times every time FW device gets one instruction. The maxx bandwidth can only be utilised if the device always knows what to do. That is not always the case, because of the buffering of instructions. Yes, an USB2 drive can excel a FW400 drive if neither drive are not lagging on instructions, but "in reality" FW400 outperforms USB2 drive almost always -- which proves my point. Want to talk about buffers? Fine. The buffers exist for a reason, which is this.

The reason why SATA/eSATA is at least as fast as SCSI drives connected via PCI is because they are connected to even wider buss than PCI. Actually SATA is about ten times faster than PCI, and because eSATA should be connected internally to the SATA buss, it's just as fast as internal SATA. But that doesn't make the earlier analogy invalid; it's just one proof to the point.
 
That's not a legitimate comparison. You're comparing the full path of one to just a plugin on another. TDM systems don't have 1 or 2 samples of latency total, there's more caused by the converters themselves and probably some more caused by the hardware before plugins are even added. Find me numbers on those (I haven't found HD, but found a reference saying the previous PT hardware was 80 samples, slower than the 64 in Logic's lowest setting) and we'll compare. I'm a little suspicious of Digi not giving any specs on that, usually claiming either zero or very low latency.

Yes, this is a legitimate comparison. Full path one-to-one, that's right. Full path means from analog input to analog output, which means, AD conversion first, then TDM plugins, then host plugins, then DA conversion. In case of native systems the tdm part is naturally omitted, but this is what you may not have understood fully: native systems add native buffer AT ALL TIMES. It is always there, even if you didn't use any plugins at all. But if you had a PTHD session that only has TDM plugins, then PTHD does not add the native buffer to the equation. And when PTHD adds TDM latency compensation, it only compensates to the maximum latency. In other words, if you have two TDM reverbs that have 2-sample latency, then your TDM latency is 4 samples and the full path from analog to analog is ADDA-conversion plus 4 samples. The same situation with native systems is ADDA-conversion plus buffer, which is at least two times 32 samples.

I'm skeptical about the zero, and the "something" doesn't really tell me anything, particularly since different converters don't have the same latency (meaning the digi stuff may have higher converter latency than other available options). Any source on the claim that other than the converters themselves, PT has zero latency?

Read the Protools Reference Guide if you don't trust my word from it. I'm sure +800 pages is a piece of cake for someone who pays as much attention to details as you do. That document clearly states that Protools HD latency comes from AD, TDM, RTAS and DA, and that RTAS latency can be omitted. Every system must have AD/DA latency, because human ear does not hear digital signals. That's why conversion can be left out from the equation. And when we do, we're left out with TDM+Native latencies, and as I said, TDM latency can be zero (without plugins) or close to zero (with plugins) but Native latency is always the size of native processing buffer.

TDM does not require buffer, because the system itself knows how many samples it takes to process signal with any given plugin. And because of this, the maximum latency is the sum of latencies reported by plugins. If I use two TDM reverbs that report 2 samples latency, then it is 4 samples total TDM latency. But if I use two RTAS (native) reverbs while using 64 buffer, then the latency is 64 and I can only hope that the system can keep up with it. That's the big difference.
 
Yes, this is a legitimate comparison. Full path one-to-one, that's right. Full path means from analog input to analog output, which means, AD conversion first, then TDM plugins, then host plugins, then DA conversion. In case of native systems the tdm part is naturally omitted, but this is what you may not have understood fully: native systems add native buffer AT ALL TIMES. It is always there, even if you didn't use any plugins at all. But if you had a PTHD session that only has TDM plugins, then PTHD does not add the native buffer to the equation. And when PTHD adds TDM latency compensation, it only compensates to the maximum latency. In other words, if you have two TDM reverbs that have 2-sample latency, then your TDM latency is 4 samples and the full path from analog to analog is ADDA-conversion plus 4 samples. The same situation with native systems is ADDA-conversion plus buffer, which is at least two times 32 samples.

I think you misunderstand my beef. I just don't agree with the notion that the ADDA delay in a TDM system is low enough that it is either "zero" or small enough to be ignored. A comparison of converter to converter on one system should be the same on the other system, which the other comparison wasn't. Also, all ADDA delays aren't the same, which is why I don't agree with simply ignoring them.


Read the Protools Reference Guide if you don't trust my word from it.

Page? I'm also skeptical of digi making a claim like that, I'd be more inclined to believe someone who tested and verified it.
 
Read the Protools Reference Guide if you don't trust my word from it. I'm sure +800 pages is a piece of cake for someone who pays as much attention to details as you do. That document clearly states that Protools HD latency comes from AD, TDM, RTAS and DA, and that RTAS latency can be omitted.

I just checked out the doc, and I didn't see that anywhere. Are you sure you're not making assumptions?
 
I just checked out the doc, and I didn't see that anywhere. Are you sure you're not making assumptions?

I agree that it isn't very clearly documented, but the information is there anyway. I'm definetely not making assumptions but instead am using the Protools HD software on a daily basis. I know for sure what I'm talking about.
 
I agree that it isn't very clearly documented, but the information is there anyway. I'm definetely not making assumptions but instead am using the Protools HD software on a daily basis. I know for sure what I'm talking about.

Page? I've been through every mention of "latency" in that doc and it doesn't say what you claim.
 
Page? I've been through every mention of "latency" in that doc and it doesn't say what you claim.

As I said, it is not very clearly laid out for a Protools newbie. There is not a single chart that would state the obvious, but instead you need to know a little bit more about how Protools HD works. Let me explain.

First of all, the TDM processing happens on PCI cards, which is obvious. Now if you think about this a second more, you notice that if all processing happens on a card and dedicated DSP, how on earth can you process the signal with the host processor? Well, Protools has a concept of "voices", which means i/o point of the TDM mixer. Every time a signal goes in or out of TDM mixer, it eats one voice. For example, if you record one track, the system does not require a voice for AD or DA conversion but instead for streaming the signal to hard drive. If you have the same track via Aux channel for monitoring purposes (an external click track for example), it does not need a voice, because it will not be recorded.

Ok, now that you know what is a voice, then you could take a closer look to that reference guide. When it talks about inserting RTAS plugins (host processing), it specifies voice consumption for each track type and tells you how to insert RTAS plugins with minimal voice consumption. Why? Because every time the system uses a voice, there is a *buffer* involved. There is one buffer for disk streaming and another buffer for host processing, and both settings can be changed from the same preference window.

Now this is what hasn't been explained very clearly: Protools HD only uses buffer where necessary. If you stay in the TDM land, then Protools is not using any buffer at all. If you on the other hand want to mix plugin types like TDM/RTAS/TDM/RTAS, you will be needing the host processing buffer multiple times. First buffer visit (voice consumption) comes from the first RTAS instance and going back to TDM processing, and second buffer visit (voice consumption) comes from the second RTAS instance and going back to TDM mixer. So in this scenario, which for a single mono track (1 voice) would cost 4 extra voices, the Protools mixer would have to use the playback buffer two times, which doubles the host processing latency compared to only using RTAS once. Native DAW:s only have native processing, so it is easier to understand that single one-size-fits-all buffer is enough for everything.

Now you should understand. The point being, as long as you stay inside TDM mixer, everything happens in an instant (without buffers). Every TDM plugin reports how many samples it takes to process, so the total latency can be easily calculated and compensated. The system is deterministic and knows it can handle the load even with all DSP chips fully loaded. Host processing on the other hand, using that is a question of faith. The host buffer probably is enough for even heavy tasks, but it is not necessarily enough in all situations. RTAS plugin can cause a hiccup, but TDM plug never will. If you have something on the TDM mixer, it is guaranteed to perform as it should.

Okay?
 
I'm interested in a reference, not an explanation.

What is your source for the information that TDM hardware has zero latency other than ADDA when no plugins are used? If it doesn't say that in any of digi's literature, and you have no other source for it, it sounds like it's just an assumption you're making.

Just because digi doesn't say there's a delay doesn't mean there is none.
 
I'm interested in a reference, not an explanation. What is your source for the information that TDM hardware has zero latency other than ADDA when no plugins are used? If it doesn't say that in any of digi's literature, and you have no other source for it, it sounds like it's just an assumption you're making. Just because digi doesn't say there's a delay doesn't mean there is none.

My source of information is my Protools HD system that I use daily. Delay can be measured, you know. Before there was an Automatic Delay Compensation we actually had to calculate delays manually to be able to compensate for it. If you don't trust my word for it, please, ask someone else that has used Protools HD before the ADC (pre-6.7 software). The fact remain, when you calculate delay without plugins and with tdm plugins, the difference is exactly what Protools reports for those plugins.

Protools session setup reports sample-accurate round-trip total latency from analog input to analog output. It is almost always less than 128 samples with plugins loaded. This delay has been manually verified many times by many Protools engineers. (With native processing, the total delay is naturally a lot more, that's why I try to avoid native processing whenever possible). So you have plenty of other people to ask if I'm not good enough for you ;)

I'm sure you didn't mean that. And yes, it would be nice if this information was clearly explained in reference documentation. This information would be handy when explaining the situation to Protools newbies...
 
My source of information is my Protools HD system that I use daily. Delay can be measured, you know. Before there was an Automatic Delay Compensation we actually had to calculate delays manually to be able to compensate for it. If you don't trust my word for it, please, ask someone else that has used Protools HD before the ADC (pre-6.7 software). The fact remain, when you calculate delay without plugins and with tdm plugins, the difference is exactly what Protools reports for those plugins.

Protools session setup reports sample-accurate round-trip total latency from analog input to analog output. It is almost always less than 128 samples with plugins loaded. This delay has been manually verified many times by many Protools engineers. (With native processing, the total delay is naturally a lot more, that's why I try to avoid native processing whenever possible). So you have plenty of other people to ask if I'm not good enough for you ;)

I'm sure you didn't mean that. And yes, it would be nice if this information was clearly explained in reference documentation. This information would be handy when explaining the situation to Protools newbies...

How did you measure that there's zero delay added between the converters when no plugins are enabled? You're not actually saying that your statement is based on it SEEMING like there's no delay? Did you actually measure or does the session setup window give you a number?

If "delay has been manually verified many times by many Protools engineers" could you point me to one who has done so online?


UPDATE: Are you talking about the "system delay" in the system setup window? According to the manual, it shows "When Delay Compensation is enabled, this display indicates the total amount of delay (in sam-
ples) applied by the system to compensate for plug-ins and mixer routing delays." Doesn't sound like it measures the total amount of delay, and it doesn't imply that no delay is added between the converters.
 
Did you actually measure or does the session setup window give you a number?

Yes, I have measured it. Numerous times. Delays without plugins and with plugins, and comparing that difference to whatever those plugins report in the mix window. They match. Back in the day you *had to* measure those delays if you wanted to put in a phase coherent mix. Now we don't have to do that anymore, thank God.

UPDATE: Are you talking about the "system delay" in the system setup window? According to the manual, it shows "When Delay Compensation is enabled, this display indicates the total amount of delay (in samples) applied by the system to compensate for plug-ins and mixer routing delays." Doesn't sound like it measures the total amount of delay, and it doesn't imply that no delay is added between the converters.

No. That seems to be the buffer size setup. The one where you can set all other buffers as well. It only says how many samples is possible to compensate. When you open the session setup window (the one which shows time code settings and such), there you have the full round latency in samples. It changes according to what plugins you have instantiated. And it's *the* value to be compensated if audio is fed to an external video device for example.

Why is it so hard to accept that TDM systems have lower latency than Native systems?
 
Yes, I have measured it. Numerous times. Delays without plugins and with plugins, and comparing that difference to whatever those plugins report in the mix window. They match. Back in the day you *had to* measure those delays if you wanted to put in a phase coherent mix. Now we don't have to do that anymore, thank God.

How did you measure it?

No. That seems to be the buffer size setup. The one where you can set all other buffers as well. It only says how many samples is possible to compensate. When you open the session setup window (the one which shows time code settings and such), there you have the full round latency in samples. It changes according to what plugins you have instantiated. And it's *the* value to be compensated if audio is fed to an external video device for example.

I don't see it in that window (for reference, the window is shown on page 795 of the latest reference guide). It just doesn't seem to be there, where do you see it in that window?

Why is it so hard to accept that TDM systems have lower latency than Native systems?

That's not what I don't accept. I just don't buy that there's zero latency other than the converters. And you haven't really made the case for that beyond "I use it every day" and "it's in the manual" which now you seem to admit it's not.
 
How did you measure it?

Scenario1: audio track without plugins, routed externally through DA/AD conversion and a loopback cable. send one-sample spike and record it back to another channel. nudge back X samples to align to the original.

Scenario2: audio track with plugin, routed externally through DA/AD conversion and a loopback cable. send one-sample spike (through the plugin) and record it back to another channel. nudge back Y samples to align to the original.

This is how it has always been done. Then calculate Y-X to see how many samples the plugin needs for processing. Nowadays the plugin reports the needed delay and there is ADC to help us compensate. But just to be sure, it can be verified: repeat steps one and two and do the math; surprisingly, the result matches to whatever the plugin reports.

Now repeat this same procedure when using RTAS native processing, and not only does the native buffer repsesent itself, but depending on the plugin there might also be additional plugin processing latency (lookahead dynamics processing, anyone).

I don't see it in that window (for reference, the window is shown on page 795 of the latest reference guide). It just doesn't seem to be there, where do you see it in that window?

I don't have a page 795 on my Protools HD 7.2 reference guide. I'm so sorry, but this time I confess that I'm too lazy to download a 7.3 guide as printed manual is so much easier to my eyes. What I mean is the Setup/Session window that can be opened with Apple-two key combination. It shows total system delay, which *can* be trusted.

For example, my current fairly large mixer reports 188 samples, which @96kHz means almost 2ms. That's a lot, but there's a lot going on in my mixer as well. But just for demonstration -- when I close that session and open a fresh one, the session setup window reports zero samples total system delay. The number grows step by step when I build my mixer.

To give you a basic real-world example; I feed one audio stream through AD and DA conversion and the session setup window reports 29 samples total delay. That is 14 samples for AD conversion, 1 sample for connecting to TDM mixer (because I use dithered mixer so there's processing involved) and 14 samples for DA conversion. Feed another stream and the window still reports 29 samples as the two naturally happen simultaneously. Insert a ReVibe reverb to the track, the plugin reports 2 samples delay and the session window now reports 31 samples total delay. Insert another ReVibe reverb to the same track, the session window reports 33 samples total delay. It all makes perfect sense and the latency does grow with whatever you build to the mixer.

So, as all systems have AD and DA conversion, those can be omitted. If we did so, let's say this basic example goes like this:

1) fresh start, zero delay
2) one stream, one sample delay
3) two streams, one sample delay
4) two streams plus one reverb (on track one), three samples delay
5) two streams plus two reverbs (both on track one), five samples delay

Need I say more?
 
Yeah, ProTools (TDM) is amazing. *sigh* I wish native systems/CoreAudio would have this kind of latency. I wonder if it's a hardware issue, and there really is a reason TDM hardware is expensive, or if it's some ingenious software design that nobody else just figured out.
 
Yeah, ProTools (TDM) is amazing. *sigh* I wish native systems/CoreAudio would have this kind of latency. I wonder if it's a hardware issue, and there really is a reason TDM hardware is expensive, or if it's some ingenious software design that nobody else just figured out.

They HAVE figured it out:
http://www.apogeedigital.com/products/symphony_performance.php

What JFreak has been saying is absolutely true; TDM systems have no additional latency added when plugins aren't in the signal path. When TDM plugs are added, there is a minimal amount of latency added. RTAS plugs are a different story.

But Apogee has created a native system that is so well integrated with the host that it can run at 1.6ms, analog to analog, through plugins. That's TDM territory. Other native systems can't touch this, especially not Firewire-based systems. FW systems HAVE TO have additional safety buffers to avoid problems at low latencies. That having been said, with a modern Intel Mac running a FW interface with well-written drivers (MOTU comes to mind), you can work pretty comfortably at 64 samples (64 in + 64 out = 128 total = approx 3ms)...I do this right now. Of course, I get added latency when I add plugins, but I get around this by direct monitoring and/or leaving plug-ins OFF until mix time. This leaves me with less flexibility than a TDM or Symphony-based system, but it gets the job done.

The Symphony system is a great proof-of-concept; native systems CAN run with latencies like TDM. But that will require a close collaboration between DAW, interface, and computer makers, and will require these people to value latency over 'frills'.
 
well, i just ambled in here in thinking tommorrow might involve me spending more money then i had planned. and i find a discussion about delay. Yay. then i see it is at a point of contention. i like points of contention. so i read.

JFreak...i remember you being a decent guy and answering questions when i was frothing at the mouth for logic. having said that, i think you are being a bit snotty in your exchange with milo.

you keep reffering to him as a newbie and you keep missing the point that he understands the theory of what you are saying. he wanted (wants) a point of reference outside of anecdote. you do not seem to have that or see the nececity. that is fine, say as much and let it be. you did good job explaining why you hold your opinion and i doubt milo would refute that. but an explanation isnt enough for everyone.

i actually find it hard believe that the delay from analog in to analog out is so small without a hard statement and a test. i get your explanation and i am not saying you are wrong. because of your explanation i am more open to the possibility of TDM being "zero latency" in certain situations. a statement from digi and a test would seal the deal.

*ApogeeGoodness*

oh my.

expresscard love.

i thought i had become stronger; that i was no longer as succeptible to gearlust. this is true. apogee raised the stakes.

bravo.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.