Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Zero Latency my ass.........help
Okay this isn't the most scientific method & someone please tell me where my experiment is wrong. I'm trying out an inexpensive ST Audio C Port Card & box for grins...........into Nuendo They claim that if use ST Audio's "mixer" software you can monitor with "zero" latency(like Total Mix,Motu's, RME's thing similar I guess). I'm not convinced so I played back an audio click that was already in the session held up the some screaming phones into an AT 4033 and recorded that track right next to the click(I should probably direct input this instead somehow but it is afterall what I'm hearing thru the phones). The two clicks are off by 12ms give or take..................every single freakin time. That's not the amount of latency they claim even if I wasn't using their propietary monitor software. Is my headphone science method screwed up? If anybody's listening I went back and output the click & brought it back in line level on to another track also & it's still off by the same amount about. My tunes at: http://www.soundclick.com/bands/5/andymostmusic.htm |
#2
|
|||
|
|||
Zero Latency my ass.........help
Phil wrote:
The 'zero latency' they refer to is only for monitoring directly from the sound-cards input to output. It's not related to any offsets incurred by driver/software timing problems. You could have a sound-card with very large buffers and 2 seconds latency, but recorded tracks should still line up, as the computer 'knows' the latency and offsets the recorded track to compensate. Okay I suppose I should figure out the offset thing. I can never get a straight answer on the latency thing..........they definitely ain't lined up It would be hell to monitor through however, It's more likely a bug in nuendo or the soundcards drivers. Is it always 12ms, or does the offset vary? Use a direct loopback to check latency offsets, as it's surprising how long sound takes to travel when you are looking at tiny amounts of time. (Though that's probably not the problem in this case, 12ms would be about 4 meters from sound source to mic I think. ). It's too long for me..just knowing it's there is bugging me. I'll try the direct looping thing. I knew it was referring to monitoring but it's a grey area with some & I figured it would line up...not. It feels good to actually see it off though & not just think it. My tunes at: http://www.soundclick.com/bands/5/andymostmusic.htm |
#3
|
|||
|
|||
Zero Latency my ass.........help
"Mondoslug1" wrote in message ... Okay this isn't the most scientific method & someone please tell me where my experiment is wrong. I'm trying out an inexpensive ST Audio C Port Card & box for grins...........into Nuendo They claim that if use ST Audio's "mixer" software you can monitor with "zero" latency(like Total Mix,Motu's, RME's thing similar I guess). I'm not convinced so I played back an audio click that was already in the session held up the some screaming phones into an AT 4033 and recorded that track right next to the click(I should probably direct input this instead somehow but it is afterall what I'm hearing thru the phones). The two clicks are off by 12ms give or take..................every single freakin time. That's not the amount of latency they claim even if I wasn't using their propietary monitor software. Is my headphone science method screwed up? The 'zero latency' they refer to is only for monitoring directly from the sound-cards input to output. It's not related to any offsets incurred by driver/software timing problems. You could have a sound-card with very large buffers and 2 seconds latency, but recorded tracks should still line up, as the computer 'knows' the latency and offsets the recorded track to compensate. It would be hell to monitor through however, It's more likely a bug in nuendo or the soundcards drivers. Is it always 12ms, or does the offset vary? Use a direct loopback to check latency offsets, as it's surprising how long sound takes to travel when you are looking at tiny amounts of time. (Though that's probably not the problem in this case, 12ms would be about 4 meters from sound source to mic I think. ). |
#4
|
|||
|
|||
Zero Latency my ass.........help
Mondoslug1 wrote:
If anybody's listening I went back and output the click & brought it back in line level on to another track also & it's still off by the same amount about. What zero latency usaually means in this context is that there is neglegable latency between your devices input and its monitor output, not that there is no latency inherent in playing out a track (which is software to output, not input to output). This matters when tracking vocalists who in some instances can be very sensitive to delay between the mic & the cans. Your software SHOULD compensate for most of the latency in the record/replay path, which as long as it can dicover how much lag is inherent in the hardware is trivial (however high the latency), but direct input-output latency is not something that can be fixed purely in software. Try this: Patch a click into input 1, and select hardware monitoring for input 1 to output 1, now patch output 1 to input 2. Arm & record both inputs, the clicks should be close to lining up... They will only be close as there is delay inherent in the AD-DA process which a monitoring solution in the digital domain will never be able to fix, but you should get within 2ms or so. To put that in perspective sound travels at about 1ft per ms in air at sea level. Regards, Dan. -- ** The email address *IS* valid, do NOT remove the spamblock And on the evening of the first day the lord said........... ..... LX 1, GO!; and there was light. |
#6
|
|||
|
|||
Zero Latency my ass.........help
Is it possible you have sampling rate missmatch from software to
hardware like you have one at 44.1/ 88.2 while the other at 48/ 96. Also if your card is internaly fixed at 48k, like SB live for instance, and you do the 44.1 recording through it, the card will hardware resample forth and back again, so your clicks may well end all over the place. You are not using SBLive, but who knows, maybe your card has the same misfortunate feature. Than again, if you are using ASIO drivers, this should (could) not happen. Are you sure you are using ASIO drivers? Maybe you are using MME drivers instead? Vladan www.geocities.com/vla_dan_l www.mp3.com/lesly , www.mp3.com/shook , www.mp3.com/lesly2 www.kunsttick.com/artists/vuskovic/indexdat.htm |
#7
|
|||
|
|||
Zero Latency my ass.........help
"Vladan" wrote in message
... Is it possible you have sampling rate missmatch from software to hardware like you have one at 44.1/ 88.2 while the other at 48/ 96. Also if your card is internaly fixed at 48k, like SB live for instance, and you do the 44.1 recording through it, the card will hardware resample forth and back again, so your clicks may well end all over the place. You are not using SBLive, but who knows, maybe your card has the same misfortunate feature. Than again, if you are using ASIO drivers, this should (could) not happen. Are you sure you are using ASIO drivers? Maybe you are using MME drivers instead? Vladan I have a SBLive card and only recently heard that it's set to 48k. I know it can cause timing problems for Cubase users, but I use Logic and I've never had any apparent timing problems recording at 44.1K, so it was a surprise to me to hear about this. This raises two questions in my mind... 1. If I change to recording at 48K, will it use less processing power and enable me to do a bit more than I can now? 2. Since the card is set to 48K, why is their "Creative Recorder" set at 44.1k? Another question... If I change to doing all my future recording at 48K, what would be the best converter to use to change the final mix to 44.1k for burning to CD? Lynn -- Listen to my music... http://www.soundclick.com/lynn http://www.soundclick.com/chaslyn http://www.soundclick.com/dickosboogieband http://www.soundclick.com/johnmckeon --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.507 / Virus Database: 304 - Release Date: 04/08/03 |
#8
|
|||
|
|||
Zero Latency my ass.........help
I have a SBLive card and only recently heard that it's set to 48k. I know it
can cause timing problems for Cubase users, but I use Logic and I've never had any apparent timing problems recording at 44.1K, so it was a surprise to me to hear about this. This raises two questions in my mind... 1. If I change to recording at 48K, will it use less processing power and enable me to do a bit more than I can now? 2. Since the card is set to 48K, why is their "Creative Recorder" set at 44.1k? Another question... If I change to doing all my future recording at 48K, what would be the best converter to use to change the final mix to 44.1k for burning to CD? You can fuss around with the SB. Or spend a relatively small amount on a card more suited to multitrack audio. A favourite choice is the M-Audio Audiophile. |
#9
|
|||
|
|||
Zero Latency my ass.........help
dmills wrote:
What zero latency usaually means in this context is that there is neglegable latency between your devices input and its monitor output, not that there is no latency inherent in playing out a track (which is software to output, not input to output). This matters when tracking vocalists who in some instances can be very sensitive to delay between the mic & the cans. I've been using an external mixer to monitor the pre-recorded tracks in Nuendo & what I'm about to overdub, this kind of takes care of the monitoring with zero latency issue...........although that's a whole 'nother argument. Sounding like a broken record I am. Your software SHOULD compensate for most of the latency in the record/replay path, which as long as it can dicover how much lag is inherent in the hardware is trivial (however high the latency), but direct input-output latency is not something I'm hoping it does but I was curious if there was an exact science to it other than trying to play back something perfectly in time with a pre-recorded track & see if they line up. The bottomline is how it sounds obviously. that can be fixed purely in software. Try this: Patch a click into input 1, and select hardware monitoring for input 1 to output 1, now patch output 1 to input 2. Arm & record both inputs, the clicks should be close to lining up... They will only be close as there is delay inherent in the AD-DA process which a monitoring solution in the digital domain will never be able to fix, but you should get within 2ms or so. I did do this and then even sent the 2nd output to the 3rd chanel input & record...........they appear to be off about 50samples each time it goes out & in...so that's cool. To put that in perspective sound travels at about 1ft per ms in air at sea level. Regards, Dan. -- ] I think i'm just overreacting........the software does appear to be compensating for the latency.........just wondered if it was dead on.............I suppose this is where rock solid drivers would come in. Thanks for the advice. ** The email address *IS* valid, do NOT remove the spamblock And on the evening of the first day the lord said........... .... LX 1, GO!; and there was light. My tunes at: http://www.soundclick.com/bands/5/andymostmusic.htm |
#11
|
|||
|
|||
Zero Latency my ass.........help
John Cafarella wrote:
"Mondoslug1" wrote in message ... Mike Rivers wrote: In article writes: The two clicks are off by 12ms give or take..................every single freakin time. That's not the amount of latency they claim even if I wasn't using their propietary monitor software. Is my headphone science method screwed up? That's a decent experiment, assuming you held the phones right up to the mic and the transmission thorugh the air was negligible. When I've tested latency, I just connect a line output to a line input and record on another track. The new math of audio interface latency is that zero doesn't equal zero. 12 ms. is kind of long though. I'd expect a bit less than 2 ms if you have everything set up correctly, but that's probably pretty difficult to do until you've spent some time learning the innards of the program. hah. Yeah, Its a work in progress. I've got an ST-audio interface. From your description it seems like you're still monitoring though Nuendo and incurring the latency there. You need to use the "External Links" program and the external mixer application to set up your zero latency monitoring. Hey John, I have been using the "External Links" and also have tried without it. Actually everything is relatively okay, my own experiment somewhat confused me. There's going to be that latency as most have pointed out to me & I suppose Nuendo is correcting that. Just that when I outputted the first click & saw it off - I freaked a bit. At least I sort of learned about what's actually going on delay-wise when you go in & out of the converters.(sort of) I fooled around with this a while ago, but found that using an external mixer was easier. I just started doing that & would agree it sounds better for monitoring, panning's better & nice to have an actual piece of hardware to deal with instead of another window to toggle back & forth with. Sorry I'm a bit light on detail here, but it was about a year ago. FWIW I've found the ST-audio interface to be absolutely rock solid. This is on an Asus CUSLC-2 Mobo under win98SE. With XP and my rig rock solid it ain't. It clicks & pops once in awhile & then it will stutter, but I'm still working on tweaking the buffer size & all that. Does alot for the price though. Thanks for the reply. -- John Cafarella End Of the Road Studio Melbourne, Australia My tunes at: http://www.soundclick.com/bands/5/andymostmusic.htm |
#12
|
|||
|
|||
Zero Latency my ass.........help
Mondoslug1 wrote:
dmills wrote: Try this: Patch a click into input 1, and select hardware monitoring for input 1 to output 1, now patch output 1 to input 2. Arm & record both inputs, the clicks should be close to lining up... They will only be close as there is delay inherent in the AD-DA process which a monitoring solution in the digital domain will never be able to fix, but you should get within 2ms or so. I did do this and then even sent the 2nd output to the 3rd chanel input & record...........they appear to be off about 50samples each time it goes out & in...so that's cool. I think i'm just overreacting........the software does appear to be compensating for the latency.........just wondered if it was dead on.............I suppose this is where rock solid drivers would come in. I suspect that that 50 sample delay is actually the delay inherent in the FIR filter in the ADC/DAC chip used as part of the oversampling logic. While it could be compensated for by having the driver report it in some way the standard driver sample position reporting mechanism is not really suitable for this. 50 samples is only a little over 1ms anyway so I wouldn't sweat it. I suspect that the real issue here is that the driver API is not designed with a mechanism to report 'fixed' delays outside of those inherent in the sample buffers. Regards, Dan. -- ** The email address *IS* valid, do NOT remove the spamblock And on the evening of the first day the lord said........... ..... LX 1, GO!; and there was light. |
#13
|
|||
|
|||
Zero Latency my ass.........help
Mike Rivers wrote:
The new math of audio interface latency is that zero doesn't equal zero. 12 ms. is kind of long though. I'd expect a bit less than 2 ms if you have everything set up correctly, but that's probably pretty difficult to do until you've spent some time learning the innards of the program. Whats wrong with a relay or fet across the audio lines? Anything post converter is going to suffer from the FIR fliters.... Regards, Dan. -- ** The email address *IS* valid, do NOT remove the spamblock And on the evening of the first day the lord said........... ..... LX 1, GO!; and there was light. |
#14
|
|||
|
|||
Zero Latency my ass.........help
|
#15
|
|||
|
|||
Zero Latency my ass.........help
Mike Rivers wrote:
In article writes: Whats wrong with a relay or fet across the audio lines? Nothing except cost, and you know how manufacturers hate that. I was thinking that something integrated into the codec should be possible. It is a real shame that there is no standard way to export this information over any of the standard multichannel digital audio formats, as some of the external converter manufacturers could take advantage of the info if it was available. Oh, and there would be that pesky little problem about making the spec sheet look a little worse since a relay would introduce a bit of noise and a FET switch would introduce a bit of distortion. Ahh yes, when you don't know any better the numbers are just so important (even when you do not know what they mean...). I know a few home studio owners like that! I think that there are hooks in the ASIO drive specification that could be used to operate it and give you monitor switching just like on a real multitrack recorder. Yes I do beleve there are. Regards, Dan. -- ** The email address *IS* valid, do NOT remove the spamblock And on the evening of the first day the lord said........... ..... LX 1, GO!; and there was light. |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
More Latency & PTHD questions | Pro Audio | |||
DAW: Monitoring the analog way | Pro Audio | |||
Tascam US 224 latency | Pro Audio | |||
So in conclusion...M Box | Pro Audio | |||
ST Audio interface opinions? (longish) | Pro Audio |