How to Edit a Podcast With Descript: Step by Step

Affiliate links ↓

Updated · May 9, 2026

You’ve got 47 minutes of raw conversation, two speakers occasionally talking over each other, and an “uh” count somewhere in the triple digits. The good news is that Descript treats audio editing like document editing — you cut words from a transcript and the audio disappears with them. Before you start: you’ll need a Descript account (the free tier gives you 1 hour of transcription per month; the Creator plan runs around $24/month for 10 hours), your audio file in MP3, WAV, or M4A format, and a rough sense of which segments you want to keep. Budget 30–45 minutes to edit a typical hour-long episode from raw recording to publish-ready file.

Step 1: Create a project and import your audio

About 5 minutes.

After logging in, click New Project from the Descript homepage. Give it a descriptive name — something like “EP 47 Raw” is easier to find than “Untitled Project 3.” Then drag your audio file directly into the project window, or click Import Media from the toolbar at the top.

Descript immediately starts transcribing on its cloud servers. For a 45-minute episode, expect 3–5 minutes of processing. You can close the tab and come back — the transcription will finish in the background.

If you recorded separate tracks for host and guest (common with remote interviews), import both files at the same time. Descript handles multi-track imports and will prompt you to label each speaker once transcription finishes.

Step 2: Assign speakers and review the transcript

About 10–15 minutes for a 45-minute episode.

Once transcription completes, you’ll see the full script view with auto-assigned speaker labels like “Speaker 1” and “Speaker 2.” Click any speaker label to rename it — “Host,” “Guest,” your actual names — whatever matches your show format. This matters later when you want to filter your edits or export individual tracks.

Skim the transcript for errors. Descript’s accuracy is genuinely good for clear recordings, but proper nouns, industry jargon, and quieter passages will have mistakes. Click directly on a wrong word and type to fix it.

Don’t get obsessive here. Fix errors in words you plan to keep; ignore errors in sections you’re about to delete anyway. The goal at this stage is orientation, not perfection.

Step 3: Cut your content by editing the text

This is your main editing pass — time varies by episode.

Read through the transcript and highlight anything you want to remove: a rambling tangent, a false start, a segment that ran long without adding anything. Highlight the text, then hit Delete. The corresponding audio is gone.

A few cuts that improve almost every podcast episode:

1. Highlight the first 20–30 seconds of pre-show small talk before the episode actually starts. Delete it.
2. Find “so, as I was saying…” or “um, yeah, so…” openers on answers. Select from the start of the sentence to where the speaker gets to the actual point. Delete.
3. Long pauses between sentences appear as gaps or blank space in the transcript view. Highlight and delete them.
4. If an entire story or anecdote didn’t land — select the whole section, listen once to confirm, delete.

If you’re unsure about a cut, right-click and choose Add Highlight to flag it and move on. Descript also offers a Scenes view that breaks your episode into chunks by topic — useful for longer shows with distinct segments. You’ll find it in the left panel under the script view.

Step 4: Kill the filler words and clean up the audio

About 2–3 minutes, mostly automated.

Go to Actions > Remove Filler Words in the top menu (in some interface versions it’s under the Edit menu or accessible via the wand icon). Descript scans the transcript for “um,” “uh,” “like,” “you know,” and similar. You’ll get a list of every instance with a checkbox next to each. Review the list — uncheck anything that sounds intentional before clicking Remove Selected.

Next, apply Studio Sound. Click on a speaker track in the timeline at the bottom of the screen, then look for the Studio Sound toggle in the inspector panel on the right side. Turn it on. Descript’s AI enhancement reduces background noise, room reverb, and microphone harshness in one step. It works well for moderately clean recordings; heavy noise can sometimes produce a slightly hollow quality, so check playback before keeping it.

Also look for Remove Long Silences under the Edit menu or timeline settings. A threshold of around 1.5 seconds works well for most interviews — any pause longer than that gets trimmed automatically.

Step 5: Add music and adjust your levels

About 5–10 minutes.

Click the + button in the timeline panel to add a new audio track. Drag your intro or outro music file here — or import royalty-free music directly into the project. Position the clip by dragging it along the timeline. Use the small fade handles at the clip edges to create smooth fade-ins and fade-outs.

For levels: right-click any audio clip and select Adjust Volume. A rough guide that works for most podcast formats — spoken audio should average around -14 to -18 dB, and music underneath speech should sit at least 20 dB lower than the voice. Descript shows a basic level meter during playback at the top right of the timeline.

Before you move on, do a full playback with headphones. Listen from start to finish. Individual cuts often sound clean in isolation but feel choppy in sequence — playback reveals pacing problems, abrupt transitions, and sections that still run too long. Fix as you go rather than noting them for later.

Step 6: Export the finished episode

About 2–5 minutes.

Go to File > Export — or click the Publish button at the top right depending on your version of Descript. Select Audio file. For format:

• MP3 at 128 kbps for voice-only episodes — standard for podcast distribution, keeps file size reasonable
• MP3 at 192 kbps if your episode has music or sound effects underneath speech
• WAV if your podcast host accepts it and you want an archival-quality master

After Descript renders and downloads the file, check that the duration matches what you expect. If the exported file is significantly shorter than your edited transcript, a section may have been accidentally deleted — go back and check the timeline for gaps.

Descript also has a direct publishing integration with Buzzsprout and a handful of other hosts under the Publish menu. If you use one of those, you can skip the export step and push straight from the app.

What to do if it doesn’t work

Transcription is full of errors. The most common cause is noisy audio — heavy background sound, a distant mic, or a phone recording. Apply Studio Sound to the audio track first, then delete and re-transcribe the file (Descript lets you re-run transcription without losing your project). For terminology that Descript consistently gets wrong, use Find and Replace to fix the same error across the whole transcript at once.

A cut sounds like a jump. When two sentences feel clean on paper but the audio edit sounds abrupt, the issue is usually tonal shift — the speaker’s pitch or pace changed around the cut point. Zoom into the timeline and find a natural breath on either side of the cut. Trimming slightly into that breath usually smooths the transition.

Studio Sound makes the voice sound hollow or metallic. Back off Studio Sound for that speaker track and experiment with individual noise reduction settings if your version exposes them separately. For persistent noise problems, export the raw audio and run it through Krisp before importing back into Descript — Krisp handles difficult noise profiles that Studio Sound struggles with.

Taking it further

Once the core edit is solid, Descript’s AI assistant (labeled Underlord in the app) is worth exploring. From the transcript, you can generate a chapter list, draft show notes, or identify the strongest 60-second clip for social — all without leaving the editor. The Clips feature creates square or vertical audiogram exports suitable for Instagram Reels or TikTok directly from your timeline.

If you also record video alongside audio, Descript handles both tracks simultaneously. The same text-based editing cuts video and audio together — delete a sentence from the transcript and both the video clip and audio clip disappear in sync. For talking-head formats or video podcasts, it’s the fastest multi-track editor we’ve used at this price point.

Frequently asked questions

How accurate is Descript’s transcription?

For clear recordings in standard English, accuracy runs around 90–95%. Technical jargon, strong accents, and audio recorded in noisy environments pull that number down. You can correct the transcript manually or use Find and Replace to fix recurring errors across the whole document.

Can I use the free plan for podcast editing?

The free tier gives you 1 hour of transcription per month, which is enough to test the workflow on a single episode. You won’t get Studio Sound or filler word removal on the free plan — both require Creator tier or above. If you publish more than one episode per month, the free tier becomes limiting quickly.

Does Descript work for video podcasts?

Yes — import a video file instead of an audio file and the workflow is identical. Edits you make to the transcript cut both the audio and video tracks simultaneously. You can export the final cut as MP4 for video distribution or as a separate MP3 audio-only file for podcast platforms.

Is there a way to undo cuts I’ve made in Descript?

Standard Ctrl+Z (or Cmd+Z on Mac) works for recent edits. Descript also keeps a full version history — click the clock icon at the top right to browse past states of your project and restore any earlier version.

Related reads

• Best AI Tools for Podcast Editing in 2026
• Descript vs Riverside for Podcast Recording (2026)
• We Tested AI Transcription on a Noisy Podcast Recording

This article contains affiliate links. If you subscribe through one, we may earn a commission at no extra cost to you. It never changes what we recommend — we only link to tools we actually use. Full disclosure.