Can you quantify that? I used my Yggdrasil OG in my home theater setup for a long time and then the Bifrost 2 and never noticed latency issues (despite perceiving myself as sensitive to that).
I did some tests out of curiosity with the Twitch audio video sync test:
TV: Vizio OLED55-H1 (120 Hz, OLED), either built-in YouTube app (gaming mode off) or YouTube website in Chrome on Windows via HDMI into an avedio HDMI switch (gaming mode on)
Monitor: Dell S2417DG (165 Hz, 1ms latency, TFT), YouTube website in Chrome on Windows
I checked whether playing the source video with VLC instead of the YouTube version in Chrome improves things and it does not.
I recorded this with my Galaxy S10+ in slow motion mode (240 FPS, not the 960 FPS super slow motion which doesn't seem to capture sound). Using a clap board I found out that my phone's camera has a base latency of 25 ms in this mode, so I compensated for that in the below measurements. Using DaVinci Resolve I advanced frame by frame in the recorded video until the Twitch provided wave form appeared on screen (lower left area), noted that timestamp, then advanced frame by frame until the sound level indicator rose above the background noise level (the app conveniently plays a frame's worth of audio when advancing frame by frame). The difference yields a frame count, multiplied by 1/240 provides the latency in seconds before compensation for the camera oddity.
If I do that with the
raw 60 FPS video from Twitch, I get one frame of latency, i.e. up to ~16.7ms, probably because the sound becomes audible at some point between those two frames. I did
not compensate for that in the below numbers. I also did not check whether the video on YouTube matches the source video from Twitch or has additional latency because YouTube makes it rather difficult to download the 1080p 60 Hz version.
Audio paths and associated latency (negative: audio ahead of video, positive: audio behind video):
PC + TV > HDMI Switch > TOSLINK > TOSLINK switch > Reclocker > Coax > Bifrost 2:
-25 ms
PC + TV > HDMI Switch > TV > TOSLINK switch > Reclocker > Coax > Bifrost 2:
20.8-41.7 ms
PC + monitor > DisplayPort > monitor headphone out:
33.3-37.5 ms
PC + monitor > USB > ZEN DAC Signature:
50-66.7 ms
PC + monitor > USB > Yggdrasil OG:
50-58.3 ms
TV with built-in speakers:
75 ms
TV > TOSLINK > Bifrost 2:
70.8ms
TV > TOSLINK switch > Reclocker > Coax > Bifrost 2:
66.7-75 ms
TV > TOSLINK switch > Reclocker > TOSLINK > Bifrost 2:
66.7-75 ms
TV > TOSLINK switch > Reclocker > Coax > Topping E30:
87.5-91.7ms
Obviously there is some variability in those numbers, with the camera frame rate of 240 FPS = ~4.2ms per frame being the smallest contributor. Also one path having a lower latency than another generally could mean the audio arrives faster or the video arrives later. If both are delayed by exactly one hour they would be perfectly in sync despite the setup being completely useless and for gaming you'd want a very low video latency.
I was actually surprised by these results, I suspected that the
TOSLINK switch and
Reclocker would add latency compared to going straight from the TV into the Bifrost 2, which in turn I expected to have higher latency than using the built-in speakers, instead those are all roughly in the same ballpark with the TV or YouTube app responsible for the most of the latency. I also would have thought that the delta sigma Topping DAC would have less latency than the multibit Bifrost 2 with its fancy filter, instead the E30 seems to have more than 10 ms of additional latency.
Perhaps the most interesting results are the first two where extracting the audio via the HDMI switch's TOSLINK out seems to eliminate more than 45 ms of latency. This switch also happens to retain the source's sample rate while the TV upsamples it to 48 kHz, which is why I have not connected the PC directly to the TV.
With the HDMI switch used as an audio extractor, the audio in the recording was perfectly in sync, so compensating for the 25 ms latency measured with the clap board the audio was leading the video by 25 ms. Only in this situation would delaying the audio help, but of course the TV's lip sync feature is useless in this case since it doesn't provide the audio. Still, routing the audio from the PC through the TV provided one of the best results. For all other scenarios, the usual audio sync option of delaying the audio would only make the problem worse since the audio is already lagging behind the image as it is.
Obviously I would love for all of these results to be significantly closer to 0 ms (especially with the built-in speakers), but even with up to 75 ms of latency I never got the impression that something is wrong despite watching a lot of YouTube videos this way every day. I'm wondering whether the Netflix or Prime Video apps perform just as bad as the YouTube app, but this is harder to test without special test content.
But it seems that Bifrost 2 and Yggdrasil are competitive with some other DACs in terms of latency, and other factors resulted in a difference of 100 ms between the best (-25 ms) and worst (+75 ms) latency I saw with it.
That is the nicer problem to have for video playback. Is the audio delay setting in milliseconds by chance or some abstract number? I'm curious by how much you have to delay the audio.
