The field-audio master stack
Adobe Podcast restores the voice, Claude tags the scene, Stable Audio writes a licensed bed. Not a Descript cut and not Suno vocals.
Who owns what
Make
The handoff — bucket in, stems out. Not DSP and not the score.
Try Make (opens in a new tab)Adobe Podcast
The restore — Enhance Speech, noise, reverb. Not music and not TTS.
Try Adobe Podcast (opens in a new tab)Claude
The scene notes — pacing, bed prompt. Not the waveform.
Try Claude (opens in a new tab)Stable Audio
The bed — instrumental, duration-matched, licensed. Not dialogue cleanup.
Try Stable Audio (opens in a new tab)
How to use this stack
- Drop the dirty interview in the bucket. Make sends the vocal to Adobe Podcast Enhance.
- Claude reads the restored transcript and writes a bed prompt. No vocals in that prompt.
- Stable Audio renders the stem. Do not also run Suno or Udio.
- This is not audio-repurpose-suite. That one cuts a clean recording in Descript.