Skip to main content

Media and standalone tools

Beyond the project pipeline, the sidebar offers standalone tools for audio, video, and document conversion jobs.

Transcribe​

The entry point for video translation: upload a video or audio file and configure the transcription:

  • Source Language (or auto-detect) and Speaker Separation for multi-speaker content.
  • Background Separation — isolate speech from music/noise before transcribing.
  • Subtitle Length / CPL (characters per line) — readability constraints applied to the generated subtitles.
  • Translation Length Limit and Dubbing Length Tolerance — how much translated subtitles may deviate from source timing, which matters once you dub.

The result opens in the video editor: review the transcription, translate to target languages, edit subtitles and timings, then export subtitles or generate dubs.

Batch Transcription​

Transcribe many audio/video files in one job — built for podcast back-catalogues and e-learning libraries:

  • Speaker Diarization and Tag Audio Events (laughter, music, etc.).
  • Text output: individual TXT files (one per audio) or a single combined file — or both.
  • Subtitle format: SRT only, VTT only (for LMS/web), both, or none.

Batch TTS​

The reverse direction: generate speech from text at scale. Pick the target language, a voice, and an accent (American, British, Australian, Indian, …), and optionally generate subtitles (SRT/VTT) alongside the audio — useful for e-learning voice-overs in many languages at once.

Lip Sync​

Re-syncs a speaker's mouth movements to match dubbed audio, taking a dubbed video from "voice-over" to natural-looking. Processing is GPU-intensive, so jobs run in the background.

OCR​

The OCR dashboard converts scanned documents into translatable text outside the normal upload flow: upload files or a whole ZIP against a project, choose the OCR model, and optionally provide a custom prompt to steer extraction (e.g. "preserve table structure"). Jobs run in the background; outputs land on the project as translatable files.

Alignment​

Builds a translation memory from existing translated document pairs — covered in TM & termbase.