Editing with AI
The AI Editing Revolution
Podcast editing used to be the biggest time sink — three hours of editing for every hour of recording was standard. AI has compressed this dramatically. Tools now remove filler words automatically, level volume, reduce background noise, detect and cut dead air, and even let you edit audio by editing text.
This chapter covers how to use these tools to produce polished episodes in a fraction of the time.
AI Editing Tools
Descript
The game-changer. Descript transcribes your audio, then lets you edit the recording by editing the text. Delete a sentence from the transcript, and the corresponding audio disappears. It's as intuitive as editing a document.
Key features: Text-based audio editing, automatic filler word removal ("um," "uh," "like," "you know"), automatic silence removal, multitrack editing, screen recording, and AI voice cloning for corrections.
Cost: Free tier available, Pro at $24/month.
Best for: Most podcasters. The text-based editing workflow is revolutionary.
Adobe Podcast (Enhance)
Upload your audio and AI improves it — removing background noise, echo, and room tone. The "Enhance Speech" feature can make a phone recording sound like it was captured in a studio.
Cost: Free basic features, included with Creative Cloud subscription.
Best for: Cleaning up audio quality, especially for guests with poor recording setups.
Auphonic
Automated audio post-production: loudness normalization, noise reduction, leveling, and filtering. Upload your file, set your targets, and it processes automatically.
Cost: Free for 2 hours/month, paid plans from $11/month.
Best for: Final mastering and loudness normalization to meet platform standards.
Kapwing and Opus Clip
AI tools that automatically identify the most engaging clips from your full episode for social media. They detect hooks, emotional peaks, and complete thoughts, then format them as vertical video clips.
Best for: Creating social media content from episodes without manual clip hunting.
The Editing Workflow
Step 1: Import and Transcribe
Import your raw audio into Descript (or your editor of choice). Let it generate a transcript. Review the transcript for any transcription errors — AI is good but not perfect with names, technical terms, and mumbled words.
Step 2: Remove the Obvious
Filler words. Descript can remove all "ums," "uhs," "likes," and "you knows" with one click. Don't remove all of them — a few filler words sound natural. Remove the distracting clusters.
Long pauses. Shorten silences longer than two seconds to about one second. Keep some natural pauses — they add rhythm and let points land.
False starts and stumbles. Find where you restarted a sentence and delete the botched version.
Off-topic tangents. If you wandered off for three minutes in a direction that doesn't serve the episode, cut it. This is the hardest edit to make because it feels like losing content. But tighter episodes are better episodes.
Step 3: Structure and Flow
Listen through the edited version. Does it flow logically? Are transitions smooth? If you rearranged any sections (possible in text-based editing), do they sound natural in the new order?
Add transitions where needed. Sometimes a simple "Now, let's talk about..." bridges two sections that were originally far apart.
Step 4: Audio Quality
Noise reduction. Apply noise reduction to remove background hum, hiss, or consistent noise. Most editors have this built in. Adobe Podcast Enhance is excellent for severe cases.
Volume leveling. Make sure the volume is consistent throughout — no sudden loud sections or whispered parts. AI leveling tools handle this automatically.
EQ and compression. If you want to get fancy, subtle EQ (boosting clarity around 3-5kHz, rolling off mud below 80Hz) and compression (reducing dynamic range) make voices sound fuller and more professional. But these are optional — many successful podcasts use minimal processing.
Step 5: Add Music and Intros
Intro/outro music. A short musical intro (10–15 seconds) and outro create a professional feel. Use royalty-free music from Epidemic Sound, Artlist, or free libraries like Free Music Archive. Never use copyrighted music without a license.
Intro voiceover. "Welcome to [show name], the podcast about [topic]. I'm [your name]." Keep it under 30 seconds. Listeners skip long intros.
Step 6: Export
Export as MP3 at 128kbps for mono (most podcasts) or 192kbps for stereo. Mono is standard for speech and produces smaller files. Apply loudness normalization to -16 LUFS for stereo or -19 LUFS for mono — these are the standards that platforms expect.
How Long Should Editing Take?
With AI tools, your editing time should be roughly one to two times your episode length. A 30-minute episode takes 30–60 minutes to edit. A 60-minute episode takes one to two hours.
If you're spending significantly longer than that, you're either over-editing (perfection isn't the goal — good enough is) or your raw recordings need improvement (better preparation and recording technique reduce editing time).
The "Good Enough" Standard
Your episodes don't need to sound like NPR productions. They need to be: clear and easy to listen to, free of distracting noises and long silences, consistent in volume, and engaging in content.
That's it. Listeners forgive imperfect audio if the content is compelling. They don't forgive boring content no matter how perfectly it's produced.
Next: getting maximum value from every episode you produce.