Key point

Recognize speech in a local audio file or MP4, then preview and edit the created script.

Steps

1. Open Tools > Media to Script and choose one audio or video file from the system picker. Audio support depends on device decoders; video must currently be MP4.
2. Wait for inspection and confirm the file and audio-track information, then tap Start Transcription.
3. Follow any preparation prompts and keep the page open through model loading, decoding, recognition, formatting and saving. The shared workspace runs one task at a time.
4. To stop, choose Cancel and confirm. The source file is not changed or deleted.
5. When complete, review the result or its full-screen preview. Tap Edit Script to open the created local script and correct it before prompting.

Options, limits and troubleshooting

• This reads speech, not burned-in subtitles or silent video frames. Clear speech and quieter background music improve usable results; overlapping speakers, dialects and specialist words need manual review.
• Video audio is read directly; extracting an intermediate audio file first is unnecessary. Recognition does not upload your source, while saved scripts follow private-sync settings.
• No audio track: choose a file with sound. Unsupported decoding: use a device-compatible file; changing its extension does not convert it. Empty text: try a short, clear spoken clip.
• If saving fails, follow script-limit or storage guidance and retry. Reaching the end of recognition does not by itself confirm a saved script.