Fixing Audio-Video Sync Issues — -itsoffset and adelay Filter

After recording, streaming, or converting, you may encounter problems like “the lip movements don’t match the audio” or “the audio is behind/ahead of the video.” FFmpeg has several approaches to fix this sync issue. This article explains the main techniques by situation.

Verified with: ffmpeg 6.1

Identifying the Direction of the Offset

Before choosing a fix, confirm which direction the offset is in.

Condition	Meaning	Fix
Audio is later than video	Sound arrives after the action	Shift audio earlier, or delay the video
Audio is earlier than video	Sound arrives before the action	Shift audio later

1. -itsoffset — Offset Input Timestamps

-itsoffset adds an offset value to the timestamps of the file specified by the next -i.

Specifying a positive value causes that stream’s start to be delayed (the stream shifts back).

When audio is later than video (align audio to video)

Input the same file twice — take video from the second input and audio from the first input (with offset).

ffmpeg -itsoffset 1.5 -i input.mp4 -i input.mp4 -map 1:v -map 0:a -c copy output.mp4

Command explanation:

-itsoffset 1.5 — Delays the timestamps of the next -i by 1.5 seconds (audio stream shifts back)
-i input.mp4 (1st) — Input with offset; audio (0:a) is taken from this
-i input.mp4 (2nd) — Normal input; video (1:v) is taken from this
-map 1:v — Video from the second input
-map 0:a — Audio from the first input (with offset applied)
-c copy — Copy without re-encoding (fast, lossless)

Note: -itsoffset applies to the immediately following -i. Order matters.

Meaning of positive and negative itsoffset values

Value	Effect
`+1.5` (positive)	Delays that stream’s start by 1.5 seconds (shifts audio back relative to video)
`-1.5` (negative)	Advances that stream’s start by 1.5 seconds (shifts audio forward relative to video)

2. adelay Filter — Delay the Audio

When audio is earlier than video, the adelay filter can delay the audio track by a specified number of milliseconds.

ffmpeg -i input.mp4 -filter:a "adelay=1500|1500" output.mp4

Command explanation:

adelay=1500|1500 — Delays the left channel by 1500ms and the right channel by 1500ms
Use the | separator to specify the delay per channel
For stereo audio, specifying both channels is standard

Example of setting different delays per channel

ffmpeg -i input.mp4 -filter:a "adelay=1000|500" output.mp4

(Left channel: 1000ms, Right channel: 500ms) — Not in CI scope (reference example)

3. -async Option — Automatic Sync Correction

-async is an option that automatically stretches or compresses audio samples to synchronize them with the video.

ffmpeg -i input.mp4 -async 1 output.mp4

-async 1 — Automatically aligns audio to video timestamps
Useful for auto-correcting small offsets, but may change pitch
Not suitable for large offsets (using -itsoffset or adelay is more accurate)

4. Difference Between aresample and async

Option/Filter	Behavior	Use case
`-async N`	Drops/pads audio based on timestamps	Automatic minor correction
`aresample` filter	Re-converts sample rate	Removing subtle sync drift, unifying sample rates
`adelay` filter	Adds silence to audio channels to delay them	Adding a fixed delay
`-itsoffset`	Offsets the timestamps themselves	Stream-level sync adjustment

General usage guide:

Known fixed offset (e.g., 1.5 seconds behind) → -itsoffset or adelay
Small fluctuations or drift → -async 1 or aresample
Want to avoid re-encoding → -itsoffset + -c copy

Common Notes

When using -c copy, the adelay filter cannot be used (filters require re-encoding)
If the offset is large (5 seconds or more), consider reviewing the recording settings or repairing the source file first
Determining the offset amount accurately by checking stream timestamps with ffprobe first is recommended (see How to Use ffprobe)

Tested with: ffmpeg 6.1.1 / Ubuntu 24.04 (GitHub Actions runner)
Primary sources: ffmpeg.org/ffmpeg.html / ffmpeg.org/ffmpeg-filters.html / trac.ffmpeg.org