Re[2]: latency questions

[numbernine@gmx.net]

m> At 5:19 PM -0500 1/12/06, monk wrote:
>>On Jan 12, 2006, at 5:12 PM, mech wrote:
>>
>>>I'm using Ableton Live 5.x with an Edirol FA-101 firewire 
>>>interface, and running at about 4-5ms latency.
>>
>>fascinating. how did you check the latency? i must say that it's 
>>hard for me to believe that you're getting that low latency with 
>>firewire.

m> That's according to the latency calculations within Ableton's 
m> preferences (as you set the number of bytes in the buffer, it 
m> automatically calculates the latency at your selected sample rate). 
m> So, I am making the assumption that Ableton is not lying to me.

most ASIO hosts "forget" to include the time the AD/DA converters
need, usually this is 32 samples in one direction.
In Cubase SX there is a parameter to adjust (Floating audio recording
offset), how much recorded audio should be shifted, I had to set it to
66 samples (2x33). Cubase users search for "audio loopback test" at
cubase.net but I think this should also be valid for other software.

Does anybody know if the 32samples are related to the sampling
frequency?

Matthias (L)



m> Regardless, it doesn't seem enough to be audibly discernable outside 
m> a "lab" environment.

>>do you use the FA-101 with a direct out kind of thing for your main 
>>signal? i guess i'm asking: does ALL of your signal go to the 
>>computer before you hear it, or is some of it routed back to 
>>whatever you're using to amplify/record your music?

m> As you've accurately deduced, I am *not* mixing my direct signal with 
m> the processed signal (other than as an experiment to test latency). 
m> Everything goes into the computer, floats around getting wacky, then 
m> eventually gets output to the PA.  If I ran it in parallel to a 
m> direct signal it would limit me a lot more as it would audibly call 
m> attention to what latency there is.

m> This is especially so in regard to effect routing to external 
m> processors.  Every time I run an effect loop to a hardware device, 
m> the latency accretes with every D/A/D conversion (hence, I primarily 
m> use delay-based effects on the hardware side; you're not going to 
m> notice a few extra ms of delay there).

m> Thus, the overall latency is low enough that I can compensate for it 
m> through my playing, as long as the direct output is not there to 
m> distract/confuse me.  That's merely my personal preference, however, 
m> and YMMV.

m>         --m.