Latency testing Bias and some iOS interfaces
-
Getting some interesting results, lots more tests to run, but the early verdict is that Bias itself is not latent
-
Are you experiencing latency on Bias fx?
Beside the switching issue you already know I'm not having any noticeable latencyI'm running bias fx on a 1st gen iPad Pro and iOS 9.3.5
-
Spoiler alert, if you own an iRig HD2, and your wallet is in any danger, do not read further.
So here are some results, all given in miliseconds. The interfaces tried were the StudioJack Mini from Sonoma Wire Works which I will abbreviate as "SJM" and the iRig HD2 from IK Multimedia, which I will abbreviate to "HD2"
I started with Audiobus 3 with system audio input as input and system audio output as output, to establish, what as far as I can tell would be a baseline round trip latency measurment
Audiobus 3 64 frames: SJM 14, HD2 16
Audiobus 3 256 frames: SJM 26, HD2 25BiasFX Ultralow Latency, all FX bypassed: SJM 15, HD2 19
BiasFX normal Latency, all FX bypassed: SJM 21, HD2 25
BiasFX Ultralow Latency, Amp, Drive and EQ on : SJM 15, HD2 20Bias Amp Low Latency: SJM 21 , HD2 25
I think there are a few conclusions to draw from this, assuming my testing is correct.
In low latencies, the SJM performs far better than the HD2, at 256 frames (if my desktop knowledge is anything to draw upon), whatever hidden safety buffers the HD2 uses seem to be bypassed, while their presence is most certainly felt at lower latencies.
Bias FX seems to add zero to negligible latency. Fruther testing with what often are latent fx (EQ, and an amp with a baked in impulse) do not seem to add to this latency. (if so, KILLER job with the DSP at Positive Grid!)
Bias Amp is quite a bit more latent than Bias FX
-
@salvatore-di-piazza said in Latency testing Bias and some iOS interfaces:
Are you experiencing latency on Bias fx?
Beside the switching issue you already know I'm not having any noticeable latencyI'm running bias fx on a 1st gen iPad Pro and iOS 9.3.5
Its one of those things that some see as career suicide to admit, but I have been blissfully unaffected by latency this whole time of using Bias FX
As it turns out, I'm looking at at least 15msec of Round Trip Latency
I normally play guitar through my DAW, which I tend to leave set for mixing, so a bit higher buffers than some might want, which at 256 is around 13msec, so its really not much different.
Also, the distance at which I play from the amp probably skews it a bit, perception wise. That 15msec is about 17 feet worth of distance from the speaker, so, probably not going to notice that 2msec difference from my DAW
-
Ok, I see the point of this tests now.
I wonder what stats you'd get with a scarlett 2I4 2nd gen. how do you mesaure latency on iPad? -
I'm measuring the latency by running one signal straight into the computer and running a split of that same signal into and out of the iPad then measuring the number of samples between them and dividing by 44,100
Here's the numbers I get from a 2nd Gen Scarlett
http://kailuamusicschool.com/tech/round-trip-latency-roundup/
10msec @ 128 buffers -
@salvatore-di-piazza To answer you questions :
Unless your soundcard has a DSP, it's your CPU power that will actually affect the latency more than what soundcard you have.
Settings at 192KHz/24Bits with a 32 samples buffers do not mean anything if your pc can't compute in time all the effects (you will hear craclking noises).The latency was measured as follow :
1 - send the signal from the pc to the ipad (the green track on the screen)
2 - record the output from the ipad running Bias FX (the stereo at the bottom)
3 - zoom in and mesure the time difference between both signals
4 - To be even more precise, substract the output latency from your session settings (usually shown in your soundcard asio settings) -
The usable latency is affected by the CPU and system and such. The actual latency AT a given number of buffers is determined by the soundcard.
There was a time when so many players were building their firewire interfaces with that awful Dice-II chipset. An RME at 64 buffers was around 6ms while a Dice-II was around 23ms
