Browse documentation

Generating speech captions

Create a draft from the correct language, track, and range.

Prepare the input

Open the caption-generation controls from the caption workspace. Choose the track containing dialogue and a range that includes the intended speech. Listen to that range first, checking mute and solo states.

Select the language actually spoken. Model availability depends on the selected language and the local speech services available to the app; review that state before starting.

Treat the result as a draft

Run generation and inspect the proposed cues. Correct names, specialist terms, punctuation, sentence breaks, and timing. Automatic transcription is an editing aid and can mishear clean recordings as well as noisy ones.

Apply the reviewed draft to the intended sequence. Keep or discard pending changes deliberately when leaving the workspace.

If generation is empty or unavailable

Check the selected source, range, audible dialogue, spoken language, and speech-model availability. A silent or muted range cannot produce useful captions. Avoid repeatedly regenerating the same unsuitable input.

If you already have a corrected subtitle file, importing it may be a better starting point. Review its timing against the current edit before delivery.

Keep exploring

Caption generation needs attentionSubtitle import and export