10-09-2023, 06:32 AM
(10-09-2023, 12:09 AM)Nutul Wrote:(10-08-2023, 11:57 PM)Tim Curtis Wrote: Why do you think there would be a latency issue?
Signal propagation and processing latencies exist throughout digital and analog playback pipelines but are almost always way below any threshold where they would produce audible artifacts.
TTYTT I don't know how big the delay would be, but what I focus on is the different length (in the time domain) of the 2 signal paths, the SUB one baing "longer"... then, whether such delay could be audible or not, I cannot say.
First of all, I cannot hear any delay, and the subwoofer is very „dry“ i.e., the sound from it is cut off very sharply (it is the sealed variant, which is faster than the ported version). This was important to me when I selected it, so it plays along with the loudspeakers nicely (which are very fast).
I think the delay might be audible earliest if it is more than maybe 25ms, introducing some booming, because this would add around one wave length of the signal (at 40Hz). I remember that if we look at video/audio sync, then everything lower than 50ms is no problem.
SVS advertises to connect the subwoofer to the loudspeaker signal directly (which I haven‘t done), before it is sent to the loudspeakers. If we take this as a worst-case scenario, then we have the delay that is introduced through the loudspeaker cable vs. the delay introduced by the DSP and the class-D amplifier in the subwoofer.
As far as I understand it, class-D amplifiers are pretty fast and don‘t introduce much delay, and without calculating it, I would say the signal propagation through the loudspeaker cable and through the amp might be in the same ballpark. So the interesting remaining factor is the delay introduced by the DSP.
The DSP has 3 filters, a low pass, phase correction and room gain compensation. This means 6 transformations in 25ms, an initial FFT and a final one, leaving 3ms for each operation. With a modern processor, this is absolutely no problem, and I assume they chose the CPU based on this. An M4-Cortex for instance could do an FFT in under one ms. And if they are using a specially designed DSP chip then the delay is even lower.
TL;DR: I don‘t think the signal-processing delay introduces any audible latency.
Cheers, Joachim