Skip to main content
POST
Re-translation functionality is not available to all users by default. If you wish to use this feature, please contact us at support@dubformer.ai.

Path Parameters

string
required
The unique identifier of the project to re-translate.

Request Body

array
required
Array of target language script segments for dubbing.Timing and Speaker Behavior:
  • Speaker assignments are required - Each speaker_id must correspond to a speaker in target_speakers. If not specified, the request will be rejected with validation error
  • Timings may be modified - System adjusts timings to match synthesized speech length and avoid excessive speed changes, unless use_fixed_timings: true is set
array
required
Array of target speaker configurations for dubbing.
string
default:"voiceover_with_original_track"
Audio mixing mode:
  • voiceover_only - Only voiceover without the original track
  • voiceover_with_original_track - Voiceover with the original track
  • voiceover_without_original_voice - Smart vocal removal dubbing
boolean
default:"false"
Whether to use fixed timings from the target script. By default, timings may be adjusted to fit the original video better.
string
default:"normal"
TTS volume level: very_quiet, quieter, normal, louder, very_loud.

Response

string
Unique identifier for the project.
number
Number of minutes charged for this re-translation (typically 0).
string
ISO timestamp of estimated completion time.
number
Remaining balance in minutes.
number
Total number of translations for this project.

Getting Original Script Data

Use Get Project to retrieve the original output_source_script, output_target_script, and output_target_speakers as starting points for your re-translation.

Timing Considerations

  • start_ms and end_ms should align with natural speech boundaries
  • Large timing changes may affect lip-sync quality
  • Use use_fixed_timings: true to prevent automatic timing adjustments

Voice Selection

  • Use specific voice keys from Get Voices for consistent results
  • soundalike voices attempt to match original speaker characteristics
  • emotional_transfer provides more expression but may be less stable