How to Edit a Podcast in Descript: A Realistic First-Time Workflow

Most Descript tutorials are made by people who've used it for years and forget what's confusing about it on day one. This is the realistic version, written for someone editing their first episode.

Before you start

Make sure your recording is already transcribed and uploaded. Descript can record directly, but if you're coming from Riverside, Zencastr, or a separate audio recorder, just upload the file. Drag and drop into a new project; transcription happens automatically and takes 2-5 minutes for a 60-minute file.

Step 1: Clean up the transcript first

Before you touch the audio, scroll through the transcript and:

• Fix obvious transcription errors — names, technical terms, unusual words. Click the word, hit Correct, retype.
• Set speaker labels — Descript auto-detects speakers but often mislabels. Reassign once at the top and it propagates.
• Mark sections — use scene breaks (Cmd/Ctrl+Enter) to chunk the episode into rough segments. This makes everything downstream easier.

This takes 10-15 minutes on a 60-minute episode and saves you hours later.

Step 2: Use Underlord for filler word removal

Click Underlord (right-side menu) → Remove filler words. Choose which to remove (default: um, uh, like, you know). Run it.

Review the proposed cuts before accepting. Underlord is aggressive — it'll sometimes flag intentional pauses or rhetorical "likes." Reject anything that breaks natural speech rhythm.

Step 3: Cut the obvious junk

With filler words gone, scroll through and delete:

• False starts ("So I— actually, let me back up...")
• Long tangents that don't earn their time
• Technical issues (audio dropouts, mic bumps, dog barking)

Delete by selecting text and hitting Delete. The audio follows. This is the workflow that makes Descript fast — don't overthink it.

Step 4: Run Studio Sound

Select all (Cmd/Ctrl+A) → Right-click → Studio Sound. This is Descript's audio cleanup AI. On clean recordings it makes a marginal difference. On rough recordings (room echo, cheap mic, background noise) it's transformative.

Warning: Studio Sound consumes credits on lower plans. Check your remaining credits before running on a long file.

Step 5: Export

Click Publish → Export. For podcast distribution, choose:

• Format: WAV (for highest quality) or MP3 (for smaller file size)
• Sample rate: 44.1 kHz
• Bitrate: 192 kbps (MP3) is the sweet spot for spoken word

If you're also publishing video, export a separate video file at 1080p.

What to skip on your first edit

• Overdub voice cloning — overkill for a first episode, expensive in credits
• Eye Contact AI — only useful for video and only on certain plans
• Multitrack editing — unless you have layered B-roll, the single sequence editor is fine
• Custom templates — set these up after you've edited 3-5 episodes and know your style

Realistic time expectations

A clean 60-minute conversation episode should take 60-90 minutes to edit your first time. By episode 5, you'll be at 30-45 minutes. By episode 10, sometimes 20 minutes for light edits.

The speed gain over timeline editors (Premiere, Audition) compounds the more you do it.

For more on whether Descript is the right tool for you, read our Descript review. If you're comparing it to other workflows, see our Descript vs CapCut comparison.