For

Vyra for podcasters: edit real footage by describing the edit

Vyra reads the full transcript and camera feeds of an episode, then builds captioned vertical clips and a cleaned long edit from a plain request.

By Sulan Zhang, co-founder of Vyra · Updated 2026-09-24

A one-hour episode has five good clips in it somewhere. Vyra transcribes the whole recording, so you can ask "find the five most quotable 45-second moments and make vertical clips with captions" instead of scrubbing for them.

A real week with Vyra

Two hosts, one 62-minute recording, two camera feeds. She uploaded both feeds Monday night and asked for four vertical clips.

Find the four strongest self-contained moments under 45 seconds. Cut each as 9:16, show whoever is talking, switch cameras on speaker change. Word-by-word captions in bold white. Add a small lower third with each speaker's name the first time they appear in every clip. List the timestamps you picked.

It came back with four clips between 41 and 45 seconds, names on screen, and the timestamp list. One clip started mid sentence. "Start clip 3 two seconds earlier" fixed it. Tuesday she ran the filler and pause pass on the full episode and got a change list back. Wednesday all four clips went out on Shorts and TikTok, and the episode went to YouTube at 1440p.

What you make

Vertical clips for Shorts, Reels, and TikTok
Cleaned full episodes
Guest highlight montages
Explainer segments cut from the conversation

Your three most-used prompts

Find clips

Find the five strongest self-contained moments under 60 seconds where the guest makes a clear point or tells a story. List them with timestamps and a one-line summary before cutting anything.

Build a clip

Make a 9:16 clip from 14:20 to 15:05. Show whoever is speaking, switch cameras on speaker change, not on a timer. Word-by-word captions, bold white with a thin black outline, centered above the lower third. Title at the top for the first 3 seconds, "why most creators quit at month three".

Clean the full episode

Remove ums, uhs, and pauses over one second across the whole episode. Keep natural breathing room. Don't cut anything between 32:00 and 33:10, that silence is on purpose. List every cut with timestamps.

A typical workflow

Upload the recording, or each camera feed if you have more than one.
Ask for a list of candidate moments before cutting. Approve or swap.
Build clips one at a time with your caption and camera rules.
Clean the full episode with thresholds for pauses and filler.
Export clips at 1080x1920 and the episode at up to 4K on AI Pro.

What Vyra does that matters for you

Transcript search across the whole episode. "Every time we mention pricing" is one prompt.
Camera switching on speaker change when you have multiple feeds.
Captions with names. "Add a lower third with the guest's name the first time they speak."
A timestamped change list on request, so you can check a cleanup without re-listening.

What it does not do

It doesn't record or host the podcast.
It doesn't rescue bad audio beyond basic levels. Record well.

Example

Example creator (TODO)

FAQ

Can it handle two or three camera angles? Yes. Upload each feed and tell it how to switch. On speaker change, on reaction, or a rule you give.

How accurate is the transcript? Accurate enough to search and caption. Fix individual words by timestamp when needed.

Can all clips match one style? Yes. Describe the style once, then say "same style" for each clip.