Existing captions, then your edits
Bring accessible captions from a public YouTube video into a text draft. Check the words against the video, edit them and keep a copy or TXT export.
Existing captions → editable text
Checking availability…
No accessible captions? Upload an MP4 recording to transcribe its audio. File transcription uses your free or Pro Cloud Minutes based on the audio duration.
Upload an MP4 to transcribe →Know what the source contains
This workflow requests existing YouTube captions through the caption service. Those captions may have been written by a person or automatically generated by YouTube. VoxText does not re-recognize the audio, switch to AI generation when captions are missing, or improve the accuracy of the source captions by importing them.
Use a publicly accessible video you are authorized to process. A recognizable URL is only an input check: private, restricted, removed or unavailable videos may not return captions. An ongoing livestream is outside the supported use case, even when its address has a recognized video-ID shape.
The result is plain text for review. It does not read words from video images with OCR, summarize the video, automatically label speakers, translate the captions, or generate synchronized timestamps and subtitle files.
Step by step
Open the public video yourself and check whether captions are available. Paste a complete HTTPS watch, youtu.be, Shorts or embed video link. The youtube.com, www.youtube.com and m.youtube.com hosts are recognized. Channel and playlist-only links are not individual video inputs.
Keep the page open while a request is pending. The result may take more than one check. If it is still pending, check the existing import rather than repeatedly starting new videos. Compare names, figures and important quotations with the video before editing.
Edit the current text, then use Copy or Download TXT. Copy uses the clipboard when permitted; if it fails, select the text and copy manually. TXT export uses UTF-8 plain text, and your browser handles the download. Save a copy before leaving the page or the temporary result expires.
A look at the result
Try a punctuation edit in this illustrative text. It was written for this example and is not imported from a YouTube video.
Before you import
The illustrative example adds sentence punctuation for readability. For a real quote, listen to the original video and verify the wording before citing it. Add speaker names or headings yourself only where the source supports them.
Use a complete HTTPS URL on an accepted YouTube host, with one video ID. Short links and the watch, Shorts, embed and /live/ video-ID address shapes are recognized. A /live/ path does not establish that ongoing livestream captions can be imported. HTTP links, channel pages and playlist-only URLs are outside this parser.
The import stops instead of generating a new transcript. Check the public video and its caption availability. For a recording you are authorized to process, use the existing audio workflow with MP3, M4A, WAV or MP4; that separate speech-recognition workflow uses Cloud Minutes.
Caption imports have a separate limited allowance and request limits. Your audio balance is not the caption-import allowance, and refreshing does not reset it. A missing-caption or failed request may use allowance. If a limit is reached, wait rather than repeatedly starting new requests.
Availability depends on the service and supplier. An interruption does not prove that no request was sent. Check the same link in the same browser to recover its existing import while it is retained; the service does not automatically restart a failed import.
Keep checking the same pending import when instructed. A result-too-large error is distinct from missing captions: try a shorter caption source rather than changing a filename. There is no guaranteed completion time or promise that every language and video length will return text.
Temporary result retrieval lasts up to one hour from creation of the import, including any pending time. Editing, copying and checking the result do not restart that clock. Your edits belong to the current editor, not a saved account transcript history. After expiry, submitting the link can start a new request and use allowance again.
Questions
No. This native-caption workflow requests existing captions only. It does not download and re-recognize the audio when captions are unavailable, and it does not turn screen content into text.
Native caption import does not consume the audio-minute allowance, but caption access and its separate allowance are limited. It has no promised number of free YouTube imports. The separate audio workflow offers up to 10 cumulative guest minutes and up to 30 free-account lifetime minutes in total, including linked guest use; those minutes do not renew monthly.
This workflow imports the accessible caption text it receives; it does not offer automatic translation or guarantee every language. The editor exports TXT, not SRT or VTT. Automatic summaries, speaker labels and synchronized timestamps are not added.
The server sends a normalized video URL to Supadata. A temporary result is associated with the browser identity that started it; its ID alone is not a share link. Returning in another browser or after losing that identity may not recover it. See Privacy for the current processing and retention details, and save your edited text before leaving.