The audio breaks up mid-answer
Your guest is three sentences into the best answer of the interview when their voice turns to gravel, drops out for a beat, and comes back a half-step behind. You have maybe ten seconds to decide whether this clears on its own or needs a different plan, and most hosts spend those ten seconds staring at the screen.
That ten-second window is a diagnosis, not small talk. What happens next depends on whether the audio itself is broken or just catching up, and the two problems call for different moves.
Check your own line before you touch anything else
Assume the fault is yours until you prove otherwise. Look at your own upload speed, close any app quietly eating bandwidth in the background, and confirm your video is still smooth on your end. If your side is clean and the guest's is not, you have already cut the diagnosis time in half.
- Run a quick check on your own upload speed, or just watch whether your own video stays smooth while the guest's breaks up.
- Ask the guest one plain question: is anyone else on their network streaming, backing up, or downloading something large right now.
- Look at your recording software's per-track indicator if it has one. Most tools that record locally will flag which side is struggling.
- Rule out your own setup last: a screen share, a virtual background, or a second monitor can eat enough CPU to look like a network problem.
What to say instead of pushing through
Say what is happening the moment you notice it. "Your audio just broke up for a second, keep going, I will let you know if it happens again" costs you nothing and tells the guest the silence on your end is not confusion or boredom. Guests read your dead air as a bad answer if you let them.
- "Your audio just cut out, keep talking, I will jump back in when I can hear you again."
- "We are going to switch to a phone line for a second, stay on this call too if you can."
- "That last answer is still usable, but give me the last sentence again in case we need a clean version."
- "Take your time, I am watching the connection, not the clock."
When to stop pushing through and switch to backup
Give it two full sentences. If the audio has not cleared by the guest's second attempt to finish that thought, or their video freezes along with it, stop asking them to repeat themselves and move to the fallback line. A guest who repeats an answer three times gives you the flattest version of it.
| What you hear | Likely cause | What to do right now |
|---|---|---|
| Robotic, stuttering voice while the call stays connected | Bandwidth congestion, packets arriving late or out of order | Ask the guest to close other tabs and pause any download, then give it one full exchange before deciding it is not clearing |
| Clean audio, then total silence and a frozen face | The connection actually dropped, not just degraded | Stop talking, note the time code, and give it fifteen seconds to reconnect before you call the backup line |
| An echo of your own voice under the guest's answers | The guest is on speaker or laptop audio instead of headphones | Ask them to grab any wired headphones or earbuds, even an old pair |
| One word or syllable repeating on a loop | CPU overload, usually a screen share or virtual background running | Have the guest close the screen share and turn off the virtual background |
That echo problem has a buying decision behind it too. Your own monitoring pair is what lets you catch it in the first place, and the case for what to buy covers that ahead of time. In the moment, the fix is just getting the guest off their speakers.
Moving to a backup line without losing the thread
Keep your local recording running through the whole interruption, even the parts you plan to cut later. If you have the guest's mobile number, call it and record that line as your new source while the original video call stays open on mute, just for their face.
- Call the guest's cell number if you exchanged one during scheduling, and record that call as a separate backup track.
- Leave the original platform open and muted. You still get their face to cut to later.
- Note the exact time code where the trouble started. It saves your editor from hunting for it.
- Ask one short confirming question after you reconnect, you still there, go ahead, rather than restarting the whole answer.
Salvaging the take instead of reshooting it
A choppy answer rarely needs a full retake. If your recording software captures separate local tracks, the glitch usually lives in the network stream, not in the file each side is quietly saving. Pull the clean local version instead of the one you heard live, and cut only the seconds that are truly gone.
Choosing a tool that saves those separate tracks is a decision you make once, long before any call glitches. The tools that hold up on a bad connection piece covers what to look for. This piece is what you do the moment a drop happens on whatever tool you already picked.
Build the checklist once, use it on every call
Most hosts relearn this triage from panic every time a call glitches. The steps live in memory, not on a page, and memory is the first thing pressure empties out. Pod Green Room keeps the guest's backup number, their usual platform, and their connection history on the guest pipeline. The fallback plan is already written before the call starts.
That same record holds the note to send afterward: an apology for the choppy stretch, or a request to re-answer one question over email if the drop cost you something real. Pod Green Room queues that follow-up next to the recording, inside the same podcast CRM that holds the rest of the guest's file. The recovery does not depend on you remembering it.
The rhythm returns once the audio does
Once the connection stabilizes, the interview does not need a fresh start. Pick up mid-thought with a short, plain question and let the guest find their place again. The conversational habits that keep a laggy call feeling normal, covered in keeping a remote interview from feeling stilted, take over from here.