How to edit a podcast (audio or video), step by step

Editing a podcast is the same few jobs whether it's audio only or filmed, and the order matters. Cut before you polish, so you don't clean up sound you're about to delete. Set the loudness last, because everything before it changes the level.

The free tools on this site each do one job on your device, with nothing uploaded. An episode with a track for each person still needs an editor that holds them all on one timeline; which one to use is further down. Recording it in the first place (mics, cameras, a quiet room, remote guests) is covered in how to start a video podcast.

The rough cut is about structure. Listen through once, or faster, read it: the free video to text converter transcribes up to 30 minutes at a time in your browser, with a time on every paragraph, so you can skim for the tangent, the repeated story and the slow start, and cut them. Descript works this way throughout: delete a word from its transcript and it's gone from the audio or video.

The fine cut is about flow:

• Mistakes. Keep the better take. Cut from the start of the fluffed sentence to the start of the retake, on a breath or between words, never in the middle of one.
• Long pauses. The free remove silence from video tool finds and cuts them, on up to 30 minutes at a time. In a conversation, use a gentle setting, so the gap where one person hands over to the other stays. The settings are in how to remove silence from a video.
• Filler words. A few “ums” sound human; a dozen at the start of every answer don't. Silence removal doesn't catch them, because they're as loud as speech. Descript's Remove filler words (in its AI Tools panel) lists them for you, and its Avoid harsh cuts setting skips the ones it can't cut cleanly. By hand, cut only the ones that slow an answer down.
• Crosstalk. With each person on their own track, you can lower or cut the one who talks over the other, rather than losing both.

Keep what makes it a conversation: the laugh, the reaction, the beat before a punchline. Cut every gap and every breath, and two people chatting sound like a script read fast.

Noise first. Clean each voice track before anything turns it up. The free remove background noise from video tool does it on your device, on up to 60 minutes of video at a time (longer for a sound file), with Light, Medium and Strong. Use the lightest that works: too strong and voices sound underwater. More in how to remove background noise from a video.

Then the levels. One guest is always quieter. Set each track's volume so the voices match, then use a gentle compressor to narrow the gap between someone's soft and loud moments. DaVinci Resolve's Dialogue Leveler does this for speech: it rides louder parts down, lifts softer ones and lowers sounds that aren't dialogue.

Then the loudness of the whole episode, measured over its full length in LUFS (LKFS is the same scale), plus its true peak, the highest point the sound reaches. Here's what the platforms say:

In practice, master the episode to Apple's −16 with peaks under −1: that meets Apple's spec, and it's within a couple of dB of the −14 Spotify plays music at. The tools for it: Audacity's Loudness Normalization effect, where you type the LUFS you want (its guide says to make it the last step before export); Adobe Audition's Match Loudness panel (Window > Match Loudness: Scan, pick a standard in Match To, set the true peak, Run); and in DaVinci Resolve, right-click the clips and choose Normalize Audio Levels.

A video podcast is edited like an audio one, plus the cameras. The sound still leads: decide the cuts by ear, and the picture follows.

• Sync the cameras by their sound. In DaVinci Resolve, select every angle and create a multicam clip with Angle Sync set to Sound, and it lines them up by the shape of their audio. The clap at the start makes that easy.
• Switch cameras on the speaker. Cut to whoever's talking, go to the wide shot for laughter, crosstalk and reactions, and stay on a shot long enough for it to settle. DaVinci Resolve Studio's Multicam SmartSwitch makes a first pass for you, cutting to the active speaker from the sound, lip movement and shot size.
• Hide your cuts with a camera change. When you cut out a mistake, switch angle at the same moment: it reads as an edit, not a jump. With one camera, scale the shot up 10 to 20% on every other piece instead.
• Keep lips in sync after cuts. Cut every track at once, picture and all the mics, so nothing slides out of line. Keep audio linked to its video, and delete with a ripple across all tracks rather than lifting one. The free remove silence tool cuts picture and sound together, so sync holds.
• Check sync at the end, too. Play a few seconds at the start, the middle and the end. If one camera slowly drifts over a long recording, split it into sections and re-sync each on a clear sound, such as a laugh or a clap.

Want music under the opening of the video version? The add music to video tool puts your own track under it, with fades and the music lowered when someone talks.

A transcript is the quickest start on show notes: the topics, the names, the links mentioned, and a few lines on what the episode is about. The free video to text converter makes one in your browser and saves it as text or an SRT.

• YouTube chapters go in the description: the first at 0:00, at least 3, in order, each at least 10 seconds. The free YouTube timestamp generator lets you mark them as the episode plays, or from a transcript of a file up to 30 minutes, and checks those rules. More in how to add timestamps to YouTube videos.
• Apple Podcasts chapters go in the episode description, through your host's RSS feed, or in the file itself (an MP3's ID3 tags). Apple suggests at least 3, no more than 6 an hour, for episodes over 10 minutes, each at least 2 minutes long, with titles of 45 characters or fewer. If you add none, Apple Podcasts makes chapters for full and bonus episodes in English itself, marked Automatically Created; you can turn that off in Apple Podcasts Connect.

One list can serve both: make each chapter at least 2 minutes long, start at 0:00, and keep the titles short.

Your host takes one file and sends it to every app through your RSS feed. MP3 plays everywhere; AAC (an M4A) is what Apple prefers for feeds. Speech sounds fine at the bottom of Apple's range, 128 kbps stereo, and Spotify itself streams podcasts at about 96 kbit/s on most devices and 128 on its web player, so a much higher bitrate mostly makes the file bigger.

Edited the video version? The free video to MP3 converter saves its sound as an MP3 at 128 kbps, 192 kbps or 320 kbps, or as an M4A or a WAV, without uploading it.

No one publishes a standard figure, because it depends on the show. Our rule of thumb (an estimate, not a measurement): a light edit, meaning a trimmed start and end, a few cuts and the levels set, takes about as long as the episode plus one listen-through. A tight edit, with every long pause and stray filler gone and several cameras, can take two to four times the episode's length. What moves it most: how many tracks and cameras, how tight you cut, and whether you work from a transcript.

Pick by what you edit: audio only and free, Audacity (or GarageBand on a Mac); mostly cutting words, Descript; several cameras, DaVinci Resolve. Whichever you pick, the free tools here can take a single job off it: the noise, the pauses, the transcript, the chapters, the MP3.

The edited episode reaches people who already listen. Clips reach people who don't: TikTok, Reels and Shorts show them to strangers by what they watch. The plan for that is in how to grow a podcast with clips, and which moments to pick is in how to clip a podcast.

By hand, clips are a second edit: going back through the episode for the moments, cutting each to stand on its own, reframing two or three people into a vertical frame and captioning every word. That's the part Clyper does. Paste the episode's YouTube link or upload the final cut, and it reads the transcript to find the moments, cuts each on whole sentences, scores its hook, frames whoever's talking (switching as the speaker changes) and captions every word in one of 5 styles. You get up to 30 clips from an hour-long episode, as 1080 × 1920 MP4s with no watermark. More in podcast clips with Clyper.

Audio-only show? The free audiogram maker turns a moment into a video with your cover, a moving waveform and captions: see how to make an audiogram.