Every cut is decided by the microphones.

How does podswitch choose the camera angle?

It measures the level of every microphone many times a second and compares each one against its own noise floor, so a quiet speaker still registers as speaking. At each moment it asks which microphone clearly leads the others.

1. One person talking: their camera.
2. Several people talking: whoever is clearly louder on their own mic. If nobody leads for a couple of seconds and you have a wide camera, the wide shot.
3. Silence: after a pause, a listener, usually the person who spoke before, since they are the likeliest to be addressed.
4. A long answer: a short reaction shot of a listener, placed where that listener is not talking, so a single angle never runs too long.

A new voice has to hold the floor for a short moment before the edit switches, and every shot has a minimum length. That is why a quick “mm-hm” does not cause a cut. When a switch does happen, it lands a fraction of a second before the new speaker starts, which is how an editor would time it.

The decisions come only from the audio. Your video is never uploaded. Where podswitch frames a face, such as a person’s tile in a gallery-view call or a crop for a clip, your browser does it on your computer.

How does it remove mic bleed?

In a room, every microphone picks up everyone. podswitch builds a dialogue mix where each speaker’s mic is open only while that person is speaking and turned down while they are not, with smooth fades so the switch is not audible. When two people genuinely talk at once, both mics stay open.

The mix is AAC by default, which keeps downloads small, or a lossless 24-bit WAV if you prefer. It becomes the audio of your multicam clip, so the sound stays continuous across every cut.

How are the cameras and audio synced?

podswitch picks a reference recording, normally the longest camera file, and compares the sound of every other file against it to find where they line up. It checks the result at the start, middle and end of the recording and gives the sync a confidence rating.

• High confidence: applied automatically.
• Medium confidence: applied, and you are asked to check it in the preview.
• Low confidence: processing pauses and you can nudge the offset by frames before continuing.

If files were recorded on separate clocks and drift apart over a long session, podswitch tells you how much. At launch it warns about drift but does not correct it.

Can I change how often it cuts?

Yes. Every plan includes three pacing presets. Changing the preset re-runs only the decisions, which takes seconds, and never counts as a new project.

Preset values are starting points and may be tuned. Some paid plans add advanced controls for the detection and the mix gate, such as how much louder a speaker must be to take the shot.

What do I need to record?

For a multicam podcast, the best results come from a separate microphone for each speaker, recorded to its own track or file. Everything else is flexible:

• A camera and a mic per speaker, with each mic recorded into its camera.
• Cameras for picture and a multitrack recorder for the mics. You map each channel to a speaker.
• Extra cameras with no mic, such as a wide shot, added as angles.
• Remote interviews where each participant’s own file is recorded separately.
• A camera file and a separate microphone file for the same person. On MAX they are paired automatically.

If you have one video-call recording or a solo video instead, see the next section.

Video files can be MOV or MP4. Audio files can be WAV, AIFF, FLAC, M4A or MP3. Before anything uploads, podswitch checks for a channel that looks empty, variable frame rate footage and cameras with different frame rates, and tells you what it found. Our recording guide covers this in more detail.

What about video calls and solo videos?

When you start a project, you choose what you are editing: a podcast with several cameras or mics, a meeting or video-call recording, or just you talking to camera. These work on every plan.

Video-call recordings. Upload the single recording of an interview or virtual podcast from Zoom, Google Meet, Teams or a similar app. Everyone shares one sound track, so podswitch finds who is speaking by their voice (speaker detection on the audio).

• Gallery view, with each person in a tile: podswitch frames each person from their tile, and you match each face to a name. The edit and the clips then show whoever is talking.
• Speaker view: the call already follows the speaker, so podswitch leaves the picture as it is and tidies the pauses with jump cuts. On paid plans it makes clips, a transcript and captions.

Solo videos. One person and one camera, with or without a separate mic. There is nobody to cut between, so podswitch suggests jump cuts: it removes long pauses and, with a transcript, filler words such as “um” and “uh”. Clips, captions and the transcript work as they do for a podcast.

A call recording has everyone on one sound track, so there are no separate microphones to mix. For a podcast you record in person, a microphone per speaker still gives the cleanest result.

What are jump cuts?

Jump cuts take long pauses out of the full edit, cutting every camera and the dialogue mix at the same frames. They are optional: off by default for podcasts and suggested for solo videos.

• Removing pauses works on every plan.
• Removing filler words such as “um” and “uh” needs a transcript, so it is on paid plans.

What happens in your editor?

You download two files: the project file for your editor and the dialogue mix. Your editor needs to know where your video files are on your computer, and browsers never tell websites that, so podswitch asks you to paste the folder path once (a Mac path, or for Premiere Pro and DaVinci Resolve also a Windows path). Save the mix into that same folder, then import the project file.

Premiere Pro and DaVinci Resolve get a Final Cut Pro 7 XML file: a timeline with every cut already made, each camera on its own video track and the mix on the first audio track. To show another camera for a shot, trim one clip and extend the neighbouring camera’s clip. The rest of this section is about Final Cut Pro.

You get a multicam clip with one angle per camera plus an audio angle for the mix, and a timeline with every cut already made. If a file has moved, Final Cut Pro’s Relink Files command reconnects it. The Final Cut Pro import guide walks through it step by step.

How are social clips made?

On paid plans, podswitch also turns the episode into short clips. The dialogue mix is transcribed, and each word is matched to the speaker whose microphone it came from. Claude, Anthropic’s AI model, then reads the transcript and suggests the moments that work best on their own: a strong first line, one clear idea and a payoff. Each suggestion comes with its hook, the reason it was picked and a score out of 100. The score is the model’s judgement of the moment, not a forecast of views.

For each clip, podswitch removes the pauses, chooses a camera for every sentence (the speaker, with short shots of the listener when they react or reply), times the captions to the words and frames it vertical, square or landscape. You can preview it with your own files, change any shot, and download it as a timeline for your editor, with captions you can edit. More about social clips.

Paid plans also write episode notes from the transcript: title ideas, a description, YouTube chapters and show notes. Team workspaces let the owner invite people by email to share projects; the owner pays for the plan.

What does MAX add?

MAX has higher limits, shown on the pricing page, and three features of its own:

• AI background music. AI suggests three music ideas from the conversation, each with a 30-second sample. The track you pick plays quietly under the edit and the clips, dips when someone speaks, and has its own track in the editor files. Tracks carry an invisible watermark and are not exclusive to you.
• Social media kit. Thumbnails and post images for every clip and the episode, sized for YouTube, Instagram, TikTok, LinkedIn, X and podcast covers. They are made in your browser from your own video, with post text, hashtag ideas and chapters to copy.
• Camera and mic pairing. A camera file and a separate microphone file for the same person are paired automatically. The camera’s own sound is used only to line things up; the clean mic is used in the mix.

What does podswitch not do?

• It does not upload or store your video. Any framing of faces happens in your browser.
• It does not render the full edit as a video file: you finish and export it in your editor. Social clips can also be rendered as an MP4 in your own browser, from your own files (beta).
• It exports for Final Cut Pro, Premiere Pro and DaVinci Resolve. Only Final Cut Pro gets a multicam clip; the others get a cut timeline with one track per camera.
• It does not export for After Effects, Avid Media Composer or other editors.