Video Podcast Audio Out of Sync? Why It Drifts and How to Fix It

The first few minutes of the episode look fine. By the forty-minute mark the host’s words land a beat after their lips move, and by the end the guest looks badly dubbed. Video podcasts run into this constantly, because most of them record sound and picture on different devices: microphones into an interface or podcast recorder, and one to three cameras or webcams, each recording on its own.

Out of sync is not one problem. Audio that is off by the same amount from start to finish has a different cause, and a different fix, than audio that starts in sync and slowly slides away. Telling them apart takes one clap at each end of the recording, and it points you to which of the four causes below to look at first.

First, find out which kind of out of sync you have

With every camera and recorder rolling, have one person clap once, sharply, where every camera can see it. Do the same at the very end of the recording, before anything is stopped. In the editor, line up the first clap: the spike in the audio waveform should sit on the frame where the hands meet. Then jump to the end and look at the second clap.

• Both claps line up. The devices simply started at different moments. That is an offset, a fixed gap, and lining up the start fixes the whole episode. Causes one and four below.
• The start lines up but the end is off. That is drift, a gap that grows over time. Note how far off the end clap is and how long the recording ran; you will need both numbers. Causes two and three.
• Sync wanders back and forth, or jumps at one point. Usually a variable frame rate file (cause three), or a camera that stopped and restarted partway through, creating a new starting point.

For a sense of scale, broadcast research puts the point where viewers notice at roughly 45 milliseconds of sound arriving early, or 125 milliseconds late; sound that arrives before the lips move is noticed sooner. At 30 frames per second one frame lasts about 33 milliseconds, so being off by a frame or two is enough to see on a close-up.

Cause 1: the devices started at different moments

Most video podcasts press record on three or four separate devices: the audio recorder or computer, plus each camera. They never start at exactly the same instant, so every source begins at a slightly different point. The result is an offset that stays the same from the first minute to the last. It is the most common sync problem and the easiest to fix.

The fix is a sync point, something sharp that every device records at the same moment. The clap is the simplest version. Editing software can also sync automatically by matching audio waveforms: each camera records scratch audio, the rough sound from its own built-in microphone, and the software lines that up against the main recording. Premiere Pro, DaVinci Resolve, and Final Cut Pro all include a waveform-based sync option. It only works if every camera is actually recording sound, so leave the camera microphones on during the recording and mute them in the edit afterward.

Watch for cameras that stop on their own. Some cameras cap a single clip at just under 30 minutes, many stop when they overheat during long recordings, and a camera battery can run out partway through an episode. Each restart is a new starting point that has to be synced separately. Our first camera buying guide covers how to check a camera’s continuous recording limit before buying; with a camera you already own, plan a deliberate break before the limit, run it on external power, and clap again after any restart.

Cause 2: two clocks that disagree

Every digital recorder keeps time with its own internal clock, which tells it how often to measure the incoming sound: 48,000 times a second at 48 kHz, the standard sample rate for video. No two clocks run at precisely the same speed. Consumer gear can differ by tens of parts per million, which sounds negligible until it adds up. At a 20 parts-per-million difference, two devices drift apart by about 72 milliseconds every hour, roughly two frames. A 90-minute episode can start perfectly in sync and finish visibly off.

This is drift, and unlike an offset it cannot be fixed by sliding the audio over once. There are three ways to handle it, from the quickest to the most permanent:

• Stretch the audio in the edit. Line up the start clap, then stretch or squeeze the audio clip until its end clap lands on the matching frame. Setting the clip’s exact duration is more precise than typing in a speed percentage, since the speed change involved is a tiny fraction of a percent. It works, but it has to be repeated for every episode, and for every camera that drifts.
• Record the audio into the main camera too. Send a mix from the mixer or interface into the main camera’s audio input. That camera’s sound and picture share one clock, so the main angle cannot drift, and the editor gets a clean reference to sync everything else against.
• Record everything on one device. Some video switchers record every camera input and the audio to a single drive, all running on the switcher’s clock, so the sources cannot drift apart from each other.

Separately, set every device to the same sample rate, and make it 48 kHz. A recorder left at 44.1 kHz, the music CD standard, alongside cameras at 48 kHz forces the editor to convert between the two. Most editors handle the conversion correctly, but a mismatch is one more place for timing to go wrong, and 48 kHz is what video expects.

Cause 3: variable frame rate from phones, webcams, and call recordings

Dedicated cameras record a constant frame rate: exactly 30 frames every second, or 29.97, 24, or 60, depending on the setting. Most phones, and many webcam and screen- or call-recording apps, record a variable frame rate instead, raising and lowering the number of frames per second as they go to save storage or processing power. Editing software expects a constant rate, and a variable-rate file can play slightly fast in some stretches and slightly slow in others. The sync wanders, and because the error is uneven, stretching the audio does not fix it.

To check a file, a free tool such as MediaInfo shows “Frame rate mode: Variable” when it is. The fix is to convert that file to a constant frame rate before editing; free converters such as HandBrake have a constant framerate setting for exactly this. Some phone camera apps can lock a constant frame rate at recording time, which avoids the extra step. If a remote guest sends in a laptop or phone recording, check it before it goes into the edit, not after the sync starts sliding.

While you are in the camera settings, set every camera to the same frame rate. 29.97 and 30 are different settings, and so are 24 and 23.976. Matching them keeps the editor from converting between rates on every cut.

Cause 4: delay inside a single-computer setup

Some studios record everything into one computer at once: cameras through capture cards, or webcams over USB, into OBS or a similar app, with the microphones coming in through an interface. Here the picture usually arrives later than the sound, because the camera, the capture card, and the computer each take time to process video, while audio passes through almost instantly. The result is an offset built into every recording, often a few frames, and the same every time.

Because it is constant, it only has to be fixed once. Record a clap, measure how many milliseconds the sound arrives ahead of the picture, and add that much delay to the audio. OBS has a sync offset setting for each audio source, under Advanced Audio Properties, for exactly this. Check it again whenever a camera, capture card, or resolution changes, since each change can shift the delay.

What to do before the next recording

• Clap at the start and at the end, every episode. In view of every camera. The end clap is what tells you whether you have an offset or drift.
• Match the settings on every device. Every recorder at 48 kHz, every camera at the same frame rate, every camera microphone on for scratch audio.
• Know each camera’s limits. Its continuous recording limit, how long its battery lasts, and whether it overheats; clap again after any restart.
• Check phone and guest files for variable frame rate. Convert them to a constant frame rate before they go into the edit.
• On a single-computer setup, measure the delay once. Set the audio sync offset, and recheck it whenever the hardware changes.
• Mute the camera audio after syncing. Scratch audio left playing under the main microphones is a common cause of a doubled, hollow-sounding voice.

If the studio is still being planned, building sync in from the start is cheaper than fixing it per episode: one device recording the audio and the cameras, or the main camera fed a clean mix. Our post on podcast studio setup for two to four people covers microphones, inputs, and headphones in the order to buy them. A multi-seat video podcast studio, built out with cameras, lighting, and treatment, typically runs $8,000–$15,000, but most sync problems do not need new gear at all, just matched settings and the same routine every episode.

Frequently asked questions

Why does the audio only drift out of sync on long episodes?

Because drift grows with time. Two devices whose clocks differ slightly fall further apart every minute, so a 15-minute clip can look fine while a 90-minute episode ends visibly off. Clapping at the end of the recording, not just the start, is how you catch it.

Can we just fix the sync in editing every time?

An offset takes a minute to fix in the edit, and drift can be fixed by stretching the audio, but both have to be redone for every episode and every camera. Variable frame rate files are better converted before editing than fought on the timeline. Matching settings and recording a clap at both ends prevents most of the work.

Do we need timecode for a video podcast?

Usually not for two or three cameras. Timecode is a running time stamp sent from one generator to every camera and recorder, so the editor can line up every source automatically. It earns its cost in larger setups with many cameras or frequent restarts, mostly by removing the manual sync step; for a smaller studio, a clap and matched settings do the same job.

Why does the voice sound doubled or echoey after syncing?

Usually the camera’s scratch audio is still playing under the main microphones, a few milliseconds apart, which makes a voice sound hollow or doubled. Mute or remove the camera audio tracks once the clips are lined up. If the recording itself sounds echoey before any syncing, that is the room, and our post on why your podcast sounds echoey covers the order to fix it in.

Our page for podcast and creator studios has more on how we work with video podcast setups, and if your episodes keep slipping out of sync and you cannot tell why, request a free first conversation. Send a short clip and a list of what records what, and we will tell you which of the four causes it is.