Podcast video clip in vertical format: 1080x1920 by default

A podcast clip rendered with Sume's Timeline 1.0 is vertical by default: the docs say "Default output is 1080x1920 MP4." For a 16:9 clip, set output.width and output.height yourself; both must be even integers from 256 to 2160.

Adobe's Premiere 26.5 notes describe Paper Edit, which builds a rough cut from transcript sentences for interviews and podcasts. The same idea works over the API: transcribe, pick sentences, cut, render. Facts here are from Sume's Timeline 1.0 page, read 2026-10-01.

How do I find the sentences to keep?

Call Video Inspect with transcribe: true and segmentation.mode: "sentence". It returns gapless sentence segments[], shaped like caption lines, plus word timings. Probe has_audio first on a clip you are unsure about. Transcription is the billed half; probe and stills are not. Read the full options in Video Inspect.

How do I cut and render the clip?

Cut with video_trim (start plus either end or duration) or put the in-point in a Timeline slot with source_in. A trim result drops into timeline_create video[] with source_in 0. Timeline needs audio.duration_seconds (1 to 1800) and an audio spine, and for a podcast the spine can be the episode audio itself with audio.source_in as the in-point.

{
  "audio": {
    "url": "https://media.sume.com/artifacts/artf_demo/episode.wav",
    "source_in": 312,
    "duration_seconds": 45
  },
  "video": [
    {
      "source_url": "https://media.sume.com/artifacts/artf_demo/episode.mp4",
      "start": 0,
      "duration": 45,
      "source_in": 312,
      "fit": "cover"
    }
  ]
}

What does the frame size mean for a wide recording?

Each slot's fit is cover (default), contain, stretch or blur. A 16:9 podcast video in a 1080x1920 frame under cover crops the sides; contain keeps the whole picture with bars. Choose per clip, and for the wide version override output.width and output.height. Other notes on reframing are in convert horizontal video to vertical.

What does it cost?