dubbing-response-models.ts

// actions
decomposeVideo({ src: 'video-url'}): SourceBlueprint
dub({ source_transcription, target_transcription, source_audio, voice_id }): MediaReference
recomposeVideo(targetBlueprint: TargetBlueprint): MediaReference

// types
interface SourceBlueprint {
    video: MediaReference // hosted video
    background_audio: MediaReference
    utterances: Utterance[]
    speakers: Speaker[]
}

interface TargetBlueprint {
    video: MediaReference
    background_audio: MediaReference
    utterances: Utterance[]
    speakers: Speaker[]
}

interface Utterance {
  id: string
  transcription: string
  speaker_id: string
  language: string
  clip: Clip
}

interface Clip {
  media: MediaReference
  start_time: number // offset of media from beginning of video in seconds (this will always be 0 for source clips because audio will be full length of video)
  // trim gives the range of the media that we care about
  trim: {
    start: number // standardized value from 0-1
    end: number // standardized value from 0-1
  }
}

interface Speaker {
   id: string
   name: string
   voice_id: string // generated as part of video decomposition process
}

interface MediaReference {
  src: string
  type: 'video' | 'audio'
  duration: number // i need to know how long each piece of media is without loading it
}