AI Podcast Clipper
How it works

How AI Podcast Clipper Turns Podcasts Into Shorts

Generate short video highlights, then use one completed clip to create card news when enabled for your account. Review and download files and copy before publishing.

Pipeline

The processing flow

Each stage has a narrow job. That makes the output easier to understand, inspect, and improve.

  1. 1

    Upload a podcast .mp4

    The creator uploads an .mp4 file up to 900 MB. The upload is stored under a per-user S3 prefix so it can be processed without becoming public.

  2. 2

    Transcribe the conversation

    WhisperX creates word-level timing data. The timing is used later to keep captions aligned with the rendered clip.

  3. 3

    Select highlight candidates

    Gemini evaluates the transcript for self-contained Q&A or discussion moments that can survive as short-form clips.

  4. 4

    Frame the active speaker

    Active speaker detection and face-aware framing guide the 1080x1920 vertical crop or background treatment.

  5. 5

    Burn in captions

    English captions use transcript timing. Korean captions are generated for the selected run and rendered into the video.

  6. 6

    Review and download

    Generated clips appear in the dashboard so the creator can play, download, keep, delete, or rerun the result.

  7. 7

    Create card news from a completed clip

    When card news is enabled for your account, select one completed clip to generate a standalone carousel for one credit. The saved script guides the text, and a scene from the original video becomes the cover. Edit words, body-card order, theme and cover for free, then download JPG/ZIP and copy the Instagram caption and hashtags.

Best fit

When results tend to work best

The app is strongest when the source episode contains clear, self-contained conversation moments.

Upload

Accepts podcast-style .mp4 files up to 900 MB per upload.

Highlight selection

Scores transcript segments and selects 1-4 clips per upload based on the requested run settings.

Captions

Supports English or Korean captions selected per processing run.

Vertical output

Exports 1080x1920 .mp4 clips for YouTube Shorts, Instagram Reels, and TikTok review workflows.

Download

Generated clips are played and downloaded from the authenticated dashboard through signed URLs.

Card news from a completed clip

When enabled, create standalone text cards with an original-video cover for one credit per complete set. Manual editing and JPG/ZIP download are free; creators copy the caption and hashtags and post to Instagram themselves.

Frequently asked questions

Does AI Podcast Clipper publish clips automatically?
No. The app provides video clips and, when enabled, card-news images, ZIP and caption/hashtags for review and download. Creators decide what to publish and upload it themselves.
What happens when card-news generation fails?
The one-credit reservation is released if the full card-news result cannot be completed. Technical retries are free. A requested AI regeneration creates a separate paid set, while manual edits and their rerender are free. Previously completed versions remain usable during a new render or a failed regeneration.
What happens when I delete a source or card-news set?
Deleting an uploaded video also removes its clips and all derived card-news sets. Deleting one clip removes only that clip and its derived sets, including AI regenerations. Deleting one card-news set preserves the clip and separate regenerated sets. Active work must finish before deletion; accepted deletion blocks new access while file cleanup completes.
What kind of source material works best?
Conversation-heavy podcast footage with clear speakers, usable audio, and self-contained discussion moments works best.
Can I choose the caption language?
Yes. The processing run can use English or Korean captions depending on the selected setting.