Cantonese video into the editor
Drop in the video file without converting it: MP4, MOV, MKV, WEBM, AVI, 3GP or a screen recording from your phone. The sound is pulled out on our server, and every timestamp points to the same moment in the original video. If the video is online, use Import from link: public YouTube, TikTok and Vimeo videos, Dropbox and Google Drive share links and direct file links all work. Only the audio is fetched, one video per link, and several links can wait in the queue.
Interviews, vlogs and TV clips are processed faster than their running time, then opened next to a player so you can click a line and check it.
Written Cantonese or standard Chinese?
Cantonese can be written two ways: colloquial written Cantonese, with characters such as 係, 嘅, 唔, 佢 and 喺, or standard written Chinese, which is how newspapers and many TV subtitles render Cantonese speech. The draft may lean either way and can mix both in one video. Decide which style you want and adjust while reviewing; subtitles for Hong Kong viewers often use standard written Chinese, while creators on YouTube frequently prefer colloquial characters.
Hong Kong speech switches into English constantly: meeting, deadline, OK, whole phrases. Those usually come out in English. Cantonese is a mid-resource language for the model, so clear speech gives a usable draft, and slang, names and fast banter need a careful read. Mandarin passages are better transcribed with Chinese (Mandarin) selected.
Cantonese videos people transcribe
Typical uploads:
- YouTube vlogs, food and travel channels, for subtitles.
- Interviews and news clips from Hong Kong, Macau and Guangdong.
- Church services and talks from Cantonese congregations abroad.
- Family videos and interviews with grandparents in Canada, the UK, the US or Australia.
- Recorded lessons and meetings held in Cantonese.
Cantonese subtitles and exports
Export SRT or VTT for Chinese-character subtitles; YouTube Studio, Premiere Pro, DaVinci Resolve and VLC display them from the UTF-8 file. Punctuation may be sparse in the draft, so add commas and full stops before exporting.
DOCX and PDF suit printed transcripts. Speaker labels are available on paid plans and with a Minute Pack.
Jyutping, English and audio files
Jyutping romanisation is not built in; the AI prompt export can ask an assistant for it.
The transcript is written in Cantonese as it was spoken; nothing is translated. To read it in English, open the Google Translate link on the transcript, or use the AI prompt export to hand the text to ChatGPT, Claude or Gemini. For recordings without video, see the Cantonese audio to text page.