Last week I interviewed a designer in Shanghai. The guest's network glitched and the first 15 minutes were completely wasted.

Last week's livestream — the guest dropped off three times

I've been building my personal brand for nearly two years, recording at least two remote interviews a month. What I dread most isn't running out of things to say — it's the guest dropping off mid-recording, the video turning into pixel soup, and then staring down 90 minutes of footage I have no energy to edit.

I used to rely on Riverside — it records locally on each side, so even if the guest's network dies, your side stays clean. But I made a mistake: I thought recording was the hard part. Turned out, faced with 90 minutes of audio and video, I procrastinated three full weeks before editing.

Then my friend Lin Xia (she runs a podcast for female founders called "Half-Sugar Radio") told me: try Descript. Delete a word, delete the video. I was skeptical but signed up for a month.

What the two tools actually differ on

Quick terms first: "local recording" = even if the network drops, each side records cleanly; "transcription" = the tool auto-converts spoken words into a text script.

  • Riverside: Strength is recording. Captures both local and network feeds from you and your guest — even with bad guest internet, your side stays sharp. Good for livestreams, interviews, conversations.
  • Descript: Strength is editing. Auto-transcribes audio and video into a text script. Want to cut a sentence? Delete the text and the video cuts with it. Also has AI to strip out "um," "uh," and pauses.

People I know already using them:

  • My friend Chen Mo, who does SaaS reviews — three livestreams a week, all on Riverside, because the video stays stable.
  • Lin Xia from "Half-Sugar Radio" — uses Descript to edit, because her interviews run two hours and traditional editing would break her.

If your pain point is "can't record clean," pick Riverside. If it's "can't edit," pick Descript.

My replication cost: money, time, first step

Money:

  • Riverside: Free tier is enough for most. Paid is $19/month (around 140 RMB).
  • Descript: Free tier has a 1-hour limit. Membership is $24/month (around 175 RMB).

Time: Both need 1-2 hours to learn the interface. Descript's "edit video by editing text" logic takes 30 minutes to click — I spent 10 minutes hunting for the delete button my first time.

Technical barrier: Both are web login and click. If you've used CapCut (剪映), you'll be up and running in 30 minutes.

First step: Go to riverside.fm, log in with Google, hit "Record," and do a test recording by yourself — see if the video is really that sharp. Same drill for Descript: go to descript.com, log in, click "New Project," upload a 5-minute old recording, and check how accurate the transcription is (Chinese recognition is roughly 85%).

Heads up: Descript's Chinese recognition is glitchy sometimes — industry jargon and names often get butchered. Always read through the transcript yourself before editing.

What I'd tell three types of people

  • If you're just starting and haven't recorded yet: Don't buy a membership. The free tiers of both tools will cover your first 3 episodes. Once you know you'll actually keep recording, then pay.
  • If you already have 1-2 paying clients/listeners and update weekly: I'd start with a Descript monthly ($24), spend a focused month clearing your backlog and stabilizing your schedule. Once editing speed is up, consider adding Riverside for recording upgrades.
  • If you're scaling — 4+ episodes a month, working with a team: Use both — Riverside for recording, Descript for editing. Or pick based on your core pain point (recording quality vs. editing speed). Don't buy both blindly.

One last thing: tools aren't mandatory. If you only record once a month, CapCut or the free tier is enough. Don't fall for the "AI auto-edits everything" hype. Get content out first — that matters more than which tool you use.