Find the interesting parts of a long recording
Quick answer: the Audio Highlight Extractor is designed to scan a long recording - a podcast episode, but also an interview, a lecture, or a meeting recording - and mark the sections worth a closer listen, so you do not have to scrub through the whole thing by hand. Detection is a preview today; the Transcript and Subtitles modes below, sharing this same page, is live and works now.
Drop your episode here
or click to choose a file - up to 500 MB
MP3 · WAV · M4A · FLAC · OGG · OPUS · MP4 · MOV · MKV · WEBM
Paste a direct link to an audio or video file, or a page a supported downloader can read.
Not every site is supported - private, paywalled or DRM-protected episodes can never be imported this way. If a link fails, download the episode yourself and use Upload File instead.
RSS import is planned so a whole show's feed can be browsed by episode, but it is not built yet. Use Upload File to try Transcript today.
You can keep this tab open and carry on - longer episodes take a few minutes. Refreshing the page reconnects to this job.
Upload an episode above to see its transcript here.
Smart Clips is a preview
Finds moments that could work as standalone clips - hooks, quotes, insights - with a start/end you can adjust. This mode has no working backend yet, so nothing is generated or uploaded when you select it - it will switch on here the moment it ships.
Reel Maker is a preview
Turns a clip into a captioned 9:16 vertical video with a waveform and safe-zone framing. This mode has no working backend yet, so nothing is generated or uploaded when you select it - it will switch on here the moment it ships.
Audiograms is a preview
A waveform video over your cover art, sized for the feed - no video footage required. This mode has no working backend yet, so nothing is generated or uploaded when you select it - it will switch on here the moment it ships.
Summary is a preview
A short and a detailed summary plus key takeaways, generated from the transcript. This mode has no working backend yet, so nothing is generated or uploaded when you select it - it will switch on here the moment it ships.
Show Notes is a preview
Chapters with timestamps, plus a starting point for links and guest details. This mode has no working backend yet, so nothing is generated or uploaded when you select it - it will switch on here the moment it ships.
Blog Article is a preview
Restructures the transcript into a readable article outline, not a raw paste of the text. This mode has no working backend yet, so nothing is generated or uploaded when you select it - it will switch on here the moment it ships.
Silence Remover is a preview
Finds and previews silent stretches so you can shorten them without cutting natural pauses. This mode has no working backend yet, so nothing is generated or uploaded when you select it - it will switch on here the moment it ships.
No generated assets yet
Once clips, Reels or audiograms can be rendered, finished files will appear here with format, size and expiry, downloadable one at a time or as a ZIP.
What a highlight timeline will look like
A static preview of the planned output on a 40-minute recording - not a live analysis. Nothing here is generated from your file yet.
How this differs from manually trimming audio
Manually trimming a long recording starts from a blank timeline - you scrub or listen through the whole thing to find the part worth keeping, which is the slow part of repurposing a two-hour meeting or a 90-minute lecture. Highlight extraction is meant to work from the transcript instead: it proposes candidate regions first, each with a short reason, so the work becomes reviewing and adjusting a handful of suggestions rather than searching from nothing. It is not a replacement for a real audio editor - it is a way to skip the search step.
What this tool does
Once live, it will scan a long recording's transcript and mark sections likely worth a closer listen, with an adjustable start and end for each.
Accepted input
Any long-form audio or video file up to 500 MB - interviews, lectures, meetings, webinars, podcast episodes.
Expected output
A set of highlighted regions on a timeline, each with a transcript excerpt and a reason, adjustable before export.
Common use cases
Journalists
Find the usable quote in an hour-long recorded interview.
Students
Jump to the dense parts of a recorded lecture when revising.
Teams
Locate the decision points in a long meeting recording.
Podcasters
The same detection this page shares with Smart Clips, framed for a wider set of recordings.
Limitations
- Not available to run yet - this page shows the intended interface only.
- Detection will work from the transcript, so poor audio quality that hurts transcription will also hurt detection.
- Suggested regions are a starting point, not a finished edit.
Privacy and retention
- Files uploaded for Transcript today are held only long enough to transcribe them, then removed by routine cleanup.
- Nothing is uploaded for highlight extraction specifically, since that mode does not yet exist.
- Only process content you own or have permission to use.
Related podcast tools
Frequently asked questions
Is the Audio Highlight Extractor available to use right now?
Not yet. Highlight detection is shown as a preview; selecting it will not process your file. Transcript and Subtitles, on the same page, are live today.
Does this only work on podcasts?
No - it is built for any long-form audio: recorded interviews, lecture recordings, meeting recordings, webinars, as well as podcast episodes. The underlying transcript-and-detection approach does not care what the recording is called.
How is this different from manually trimming audio?
Manual trimming starts from a blank timeline - you have to listen (or scrub) to find where something worth keeping happens. Highlight extraction is meant to propose candidate regions first, from the transcript, so trimming becomes reviewing and adjusting suggestions rather than searching a long recording from nothing.
Will it tell me why a region was chosen?
That is the plan - each region is meant to come with a short reason, the same way Smart Clips explains its picks for podcasts.
What is the maximum recording length?
The same 500 MB upload limit as the rest of the studio, since detection is planned to run on the same uploaded file as Transcript.
Does highlight extraction work on my file today?
No. Only Transcript and Subtitles are connected to a working backend. Upload a file to get a timestamped transcript now; highlight extraction will use that same upload once it ships.
Last reviewed: 2026-09-08.
Start with a transcript today
Upload your recording now to get a timestamped transcript - the foundation highlight extraction will build on once it ships.
Back to the workbench