Lip-Sync Error Audio/video synchronization (lip sync) and audio delay compensation

What lip-sync error is

Lip-sync error is the offset between what you see a mouth do and when you hear the word. The measurement standard is ITU-R BT.1359-1 (1998), "Relative timing of sound and vision for broadcasting." Its sign convention is that "a positive value indicates that sound is advanced with respect to vision." So +45 ms means the audio leads the picture by 45 ms, and -125 ms means it lags by 125 ms.

From subjective tests in Japan, Switzerland and Australia on NTSC and PAL material, the recommendation reports that "detectability thresholds are about +45 ms to -125 ms and acceptability thresholds are about +90 ms to -185 ms on the average." Between the "just detectable" limits for sound leading and sound delayed there is about 170 ms of range. The recommendation's overall tolerance for sound/picture timing is that it "shall not exceed +90 ms or -185 ms."

Broadcast limits are tighter than the viewer limits

North American broadcasters do not use the ITU thresholds. The ATSC Implementation Subcommittee's finding IS-191 says the sound should never lead the video by more than 15 ms and never lag by more than 45 ms at the input to the DTV encoder, and TV Technology's account notes that these numbers are significantly tighter than ITU-R BT.1359-1. The group also set an overall end-to-end tolerance of +30 ms (audio leading) to -90 ms (audio lagging), which corresponds to the traditional NTSC rule of +1, -2 frames. The reasoning is error accumulation: every stage in a chain adds a little, so each stage has to stay well inside what a viewer would notice.

For film, BT.1359 itself notes that a half-frame tolerance at 24 fps is about 22 ms. One frame at 24 fps lasts 41.7 ms and at 60 fps lasts 16.7 ms, so a two-frame error at 60 fps is 33 ms: inside both the +45 ms early limit and the -125 ms late limit, though outside the ATSC encoder-input window of 15 ms early or 45 ms late for sound that leads.

Where home delay comes from

Video processing. A TV that applies motion smoothing, upscaling or tone mapping delays the picture while the audio races ahead, and an external soundbar or AVR then decodes and processes sound separately. Sony's soundbar guide acknowledges this: "A time gap between the sound and picture may occur depending on the sound source," and "If your TV has a function that delays the image, use it for adjustment."

Wireless links. Feasycom's codec comparison lists typical Bluetooth A2DP delays of roughly 150 to 220 ms for SBC, 120 to 180 ms for AAC, about 100 to 120 ms for aptX Classic, 32 to 40 ms for aptX Low Latency and under 50 ms for LE Audio. Set against the ITU numbers, SBC's midpoint of about 185 ms sits at the -185 ms acceptability limit, and AAC's range overlaps the -125 ms detectability limit. Feasycom says most people start to notice lip-sync problems around 70 ms, a vendor rule of thumb that is stricter than the ITU figures. LE Audio's LC3 codec delay is 12.5 ms at 10 ms frames and 11.5 ms at 7.5 ms frames according to Fora Soft, though real end-to-end delay is larger. The Bluetooth SIG says LE Audio latency is "almost an order of magnitude lower than those commonly experienced with Bluetooth Classic audio" (see Auracast and Bluetooth audio codecs).

HDMI, eARC and automatic correction

HDMI 1.3 added Auto Lip-Sync Correction, so a display can tell the source how much video latency it adds. Wikipedia's version table lists it against HDMI 1.3, and a later HDMI 2.2 (released June 25, 2025, per the same article) adds Latency Indication Protocol support for improving audio and video synchronization; see Latency Indication Protocol. How well it works depends on whether each device in the chain reports its delay honestly, and many chains still need manual adjustment, so treat automatic correction as a starting point.

Fixing it with audio-delay settings

Delay settings add time to the audio so it lands with the delayed picture. They only work in one direction: you can delay sound, not make it arrive earlier, so fixes for audio that is already late must instead reduce video processing or the wireless buffer. Sony's HT-A7000 guide shows typical controls: for TV audio, press OPTIONS, choose SYNC, and adjust "from 0 msec to 300 msec in 25 msec increments"; for another connected device, use the A/V sync option on the TV screen, with the same 0 to 300 msec range. Sony's troubleshooting page adds a trap: if TV audio lags, "set the adjustment value of the delay between the picture and sound to 0 msec if it is set to the range between 25 msec and 300 msec," because a delay setting left on can itself cause lag.

Order of attack: first switch the TV to its low-latency or game mode (see input lag), which removes video processing delay. Then check the audio path: eARC for wired bars, a codec with low latency for Bluetooth. Only then dial in the delay on the AVR or soundbar, one 25 ms step at a time, using a clip with clear speech.

Worked example

Suppose a soundbar over Bluetooth SBC adds roughly 185 ms (the midpoint of the 150 to 220 ms range). That sits at BT.1359's acceptability limit of -185 ms and well past the -125 ms detectability limit, so most viewers would notice. Switch the link to aptX Low Latency at about 36 ms and the offset falls inside the detectability window of -125 ms. On a wired eARC bar the delay is no longer a codec issue. Suppose the picture is 80 ms late because of TV processing and the bar has no delay setting, the sound would be 80 ms early, which is inside the +90 ms acceptability limit but beyond the +45 ms detectability limit, so some viewers would still notice.