Translate a Douyin video into English: the speech, the on-screen text and the caption

A Douyin (抖音, Chinese TikTok) video conveys its meaning primarily through what is spoken and what is displayed on screen. Neither can be selected or copied, and the platform provides no captions. To access the content, paste the video's share link into LinkDigest and select English. The service transcribes the speech, reads the on-screen text, and presents both in English alongside the original Chinese, along with the video title and key points.

No Douyin account or cookies are required, and no downloads are necessary. Translation adds one credit to your account, which comes with ten free credits.

One notable aspect of Douyin videos is the scarcity of text. Most are individuals addressing the camera, and the platform does not provide transcripts. Subtitles present on the screen are merely pixels. Additionally, the on-screen text includes headlines, labels, and numbers drawn directly into each frame. The link itself is also part of the on-screen text. LinkDigest cannot translate Douyin links directly; it must first read the video.

The video in question is about one-person AI startups. The final spoken line begins at 2:17, making the video approximately two and a half minutes long. It costs 22 seconds to read the video, which requires one credit. The read process is free if the video has already been processed. The translation adds another credit, bringing the total cost to two credits. The video has been read before, so the translation is free. The original Chinese text for each field remains untouched, allowing for verification of the translation.

Sample lines from the on-screen text include "900人公司" (900-person company), "二号位CTO离职" (No. 2 CTO resigns), and "使用火" (Using fire). The last line is the account's name, translated word for word, ensuring the original Chinese name remains beside the English translation. The translation process is straightforward in a browser: sign in, paste the share link, set the translation to English, and hit read.

From a code perspective, you can add translate_to to your API call. The response provides the video title and description in English, the speech as one translated text, the on-screen text line by line, and the takeaways, each linked to its Chinese source.

The translation maintains the original timestamps, allowing for synchronization between the English sentences and the moments they were spoken. To obtain the link, share the Douyin video, then copy the link displayed in Chinese text. LinkDigest can fetch this link from the line, even if it appears within Chinese text. The link is case-sensitive, so it's best to copy it rather than retype it.

The video link must be public for LinkDigest to read it; otherwise, the service will provide a clear explanation of alternative methods. Speech recognition primarily supports Mandarin; heavy dialects or loud background music can affect accuracy. Videos exceeding 24,000 characters are translated only up to that limit, with the translation note indicating the extent of the translation. The Chinese transcript is always complete, while the translation may stop at the character limit.