While Kala & I are in Summit I took some time to take some more sweep measurements of the Nest Music Theater with zero delay applied to the center channel. I referenced in the center channel buidlout of the Nest Theater that I was seeking a more certain way of setting the delay for the center channel I recently built ~ adjustments to date have been "by ear" from clicks I applied and ranged from 42msec to 85msec depending on audio path. All actually sound great but I wanted to actually measure "what's right." So, doing an FFT (fast Fourier transform) on the sweep data, one can convert frequency to time. Applying an impulse to the front end of the sweep and looking at the FFT actually allowed me to see the impulse in time on each of the mains and the center channel. The difference in these observations is the setting needed to make the center channel signals arrive at my listening position at the same time as the mains' signal. Remember that for me those delays occur not only from speaker physical positioning but from, more dominantly, delays within different equipment producing the sound from the mains as that from the center channel. For at least streaming flac files from my Linux JRiver MC streamer that is 74msec, as can be seen in the first graph. I added that exact delay via the Felston DD740 and viola... perfect synchrony as displayed in the 2nd graph where only one zero point impulse response is observed.
Math & measurements are pretty useful. I may need to update this understanding to include measurements when other sources like YouTube or phono sources are used. These use slightly different audio paths within my gear and require different processing power within them, which will impact the delay observed. Now though I have a clear scheme to measure and see the latency directly.
Update 28Aug23... I was able to measure the latency observed on the TV Optical (for mostly YouTube) and the Phono paths, using the same process as above. For TV Optical I issued the signal on HDMI and measured via the UMIK1. The delta in impulses in time was similar to that above for the Streamer Out, but different: 68msec, as shown in the 1st graph below. Setting the DD740 to 68msec delay and measuring again, the arrival at the measurement mic is simultaneous. It is possible that when more complex material is flowing through the Dolby decoder, that processing time (and therefore delay) will increase some. However, 68msec is as close to perfect as I can get statically for this audio path. For Phono I used standard stereo analog out from my mac and measure via the UMIK1. Here the results again were about the same but I fiddled a bit with the delay on the Phono audio path and settled on 63msec delay for the CC. The last two graphs show the before and after impulse responses.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.