How to Extract Text From Short Videos
Start with a public video link or its full share text. Videosays returns spoken text and timestamps without requiring a local video upload. Before submitting, choose whether you have specific videos, a Douyin collection, or a creator profile; these sources need different selection steps.
Steps
Check that the video is publicly accessible
Open the link and confirm that it leads to the video. Private, deleted, login-restricted, or region-restricted sources may be unavailable.
Copy a video link or full share text
Use Copy link in the share panel. Full share text is accepted. For a collection or profile, copy the source page rather than one video inside it.
Choose the source on the homepage
Use Video links for one or up to 100 specific videos. Use the Douyin collection or profile option to load a list first, select videos, and review the quote. Switching options preserves your pasted text. Loading a list does not create every transcription automatically.
Sign in and follow task progress
New accounts get 10 free minutes. Review your credits and submit, then follow progress in the dashboard. Processing time depends on duration, the queue, and source access; avoid resubmitting a task that is still running.
Proofread and export
Check important names, numbers, and technical terms against the source. Copy text for notes, or export TXT, SRT, or VTT. Keep the source URL alongside the result.
Single and batch links use the same transcription service
Paste one link for a single video, or separate links for a batch of up to 100. Supported public sources include YouTube, TikTok, Instagram, X, Douyin, Bilibili, Xiaohongshu, Kuaishou, and WeChat Channels.
Collections and profiles require selection
A collection preserves episode order; a profile lets you browse public posts. Select the videos you need and confirm the duration quote. Short share URLs may not reveal their source until resolved, so choose the source you copied them from.
Original captions, speech recognition, and proofreading
Automatic mode prefers available original captions and otherwise attempts speech recognition. Results retain the source language. Background music, accents, overlapping speakers, and automatic captions can introduce errors; completion does not establish word-level accuracy.
TXT, SRT, or VTT?
TXT is convenient for reading, search, and notes. SRT contains numbered subtitle cues with start and end times. VTT is a subtitle format often used by web players. Subtitle drafts may need timing and wording edits before publication.
Choose a web, CLI, API, or agent entry
Use the web for manual work, the API for application integration, and the CLI or Skill for agent workflows. These entries use the same account credits and transcription capability; analysis and summaries are subsequent work on the resulting text.
Start with a video you want to read
Paste its public link, choose the source workflow, and review the transcript against the original.
FAQ
Can I paste the full shared text?
Yes. Videosays extracts supported video URLs from share text. Do not mix a source list and individual videos in one submission.
Does it extract the title or the spoken content?
The transcript contains spoken content or available original captions. Metadata such as the title is kept separately.
Can it read words burned into video frames?
This workflow is not visual OCR. A video needs available captions or recognizable speech.
Is there a free trial?
New accounts get 10 free transcription minutes. Viewing the public transcript example requires no sign-in and consumes no credits.