This is your go-to document for proofreading dubbing scripts for AI dubbing. Follow the pointers below to learn how to proofread your dubbing file.

1. Translation

The translation must accurately reflect the original source audio. Use vocabulary and syntax, as well as adaptation, to produce a natural sounding translation. Modify the script to your liking so the final audio sounds natural using the additional commands and tricks outlined below.

1.1. Grammar and vocabulary

• The translation should be grammatically correct and use the right vocabulary, even if the source contains errors (as errors will unnecessarily complicate the translation). The syntax should flow logically.

• When you modify content or segmentation in your dub, you can "redub" to hear the changes.

  <aside> ⌨️

  Click Redub below a segment to redub that specific segment.

  </aside>

1.2. Colloquial language

• Match the speaker's colloquial language in the target dub when possible. This helps maintain natural pacing, improves lip-synchronization, and preserves the conversational tone.
• If the speaker uses casual vocabulary, use equivalent casual terms in the target dub.
• Keep repetitions, translate filler words (when used by the source speaker), and maintain false starts if it improves syncing and remains understandable. You can isolate these filler words into separate segments if that improves their timing.
  • e.g. “[And…] [And I just find that incredible.]”

1.3. Natural and organic dialogue

• The translation should sound natural and authentic. It should flow smoothly in the target language rather than sounding like a translation. The dubber should be a native speaker who avoids literal translations that don't capture the intended meaning.

• TIP: Listen to the source audio before making corrections. This helps you understand the required pace and emphasis.

  <aside> ⌨️

  You can seamlessly toggle between the original audio and the dub using the buttons below the progress bar.

  

  </aside>

• TIP: Other ways to create a natural-sounding dub include:

  • Varying sentence structure to add dynamism.
  • Modifying emphasis where sentences have the wrong intonation (see 1.5. Emphasis and intonation).
  • Breaking sentences into shorter segments to help the AI create more natural-sounding speech.
  • Using conversational expressions instead of formal language.

• If your audio presents artefacts that disturb the naturalness, try rewriting your sentence and redubbing.

1.4. Accent

• All speakers should use an accent native to the target language. If a speaker's accent retains traces from the source language, reduce the similarity setting slightly (see: Similarity). Be aware that this adjustment will affect that speaker's voice throughout the entire file.

1.5. Emphasis and intonation