How to Get a Transcript of a YouTube Video: Convert Video to Text With or Without Captions

Want to turn a YouTube video into text? The method depends on whether the video has a caption track. If it has closed captions (CC), you can copy the text for free from YouTube's built-in transcript, which you open with the "Show transcript" button. If it doesn't, you need AI speech recognition to generate a transcript from the audio instead. This guide is for students organizing course notes, marketers turning interviews into articles, and video creators who need subtitle files.

Quick Check: Does Your Video Have Captions (CC)?

Open the video in a desktop browser, expand the description box under the title, and scroll to the bottom. If you see a Show transcript button, the video has a caption track.

For the no-captions route, this guide uses Taption for the walkthrough; other speech recognition tools work on the same principle.

Why do some videos not have this button? According to YouTube's official help page on automatic captions, common reasons include: the video's audio is still being processed, the language isn't supported by automatic captions, the video is too long, the sound quality is poor or YouTube can't recognize the speech, there is a long silence at the beginning, or several people speak at the same time or several languages are spoken at once.

Method 1: Copy YouTube's Built-In Transcript for Free

If the video has captions, follow YouTube’s transcript instructions to open the transcript. The steps below explain how to select and copy text manually in a desktop browser; this is not a YouTube caption-file download feature.

• Step 1: Open the YouTube video in a desktop browser and click ...more at the end of the description box under the title to expand the full description.
• Step 2: Scroll to the Transcript section at the very bottom of the description and click Show transcript.
• Step 3: A Transcript panel opens to the right of the player (or below it). Click any line to jump to that point in the video.
• Step 4: Click ⋮ in the top-right corner of the panel and choose Toggle timestamps to keep or hide the timestamps.
• Step 5: Pause the video first so automatic transcript scrolling does not interfere with text selection. Select the transcript text you need with the mouse, then press Ctrl + C (Cmd + C on a Mac) to copy it. Paste it into a text editor and check that both the beginning and end were copied. Copy long transcripts in sections if needed.

This illustration shows the English interface, not a record of a real-video test. The labels ...more, Show transcript, and Toggle timestamps correspond to the controls described above. Pause the video before copying.

This method has three limitations:

• You can't copy from the mobile app: In the YouTube app, you can tap a transcript line to jump to it, but for now you can't select and copy the text directly. On a phone, open youtube.com in a browser instead and switch to the desktop site.
• Embedded players don't show it: A YouTube player embedded on another website doesn't display the description box, so you have to go back to the original YouTube page to open the transcript.
• Viewers can't download the caption file: YouTube doesn't give viewers a button for downloading caption files. Channel owners can download the captions of their own videos from YouTube Studio; see YouTube's help page on editing or downloading captions in YouTube Studio.

Turn Choppy Captions Into Readable Paragraphs

Automatic captions come as short fragments, one line at a time. They commonly have very little punctuation and proper nouns in lowercase, and they usually don't indicate who is speaking. In Chinese-language videos, the captions may also show Simplified characters or wording that isn't used in Taiwan, which matters if you need the text in Taiwan Traditional Chinese.

Here is an illustrative example. This short Q&A between a host and a speaker was written to show the formatting differences. It is not the output of a real video, and the TXT and SRT samples later in this guide reuse the same content. Copied captions look roughly like this:

If you only ask ChatGPT, Claude, or Gemini to "clean up this transcript," the model will often write a summary instead. To keep the full content, use this suggested prompt, which spells out the requirements and makes omissions less likely:

You should get something like this (again, an illustration):

Generative models can still drop or rewrite sentences, so check anything you plan to quote against the video. Split long transcripts into manageable sections by topic, then paste and compare each section against the original. Also, unless the original captions already label the speakers, the cleaned-up text won't have speaker names, and it isn't suitable for use as a subtitle file as is.

Method 2: Transcribe a YouTube Video Without Captions Using AI Speech Recognition

Tools that only fetch existing captions have nothing to fetch when a video has no caption track. For those videos, you need a speech recognition tool that "listens" to the audio itself and generates the text and timeline from the sound. Cloud services and the open-source OpenAI Whisper speech recognition model both work this way. Running Whisper on your own computer means installing and configuring it from the command line, and it has no built-in speaker separation (diarization).

Taption Walkthrough: Paste the Link, Transcribe, Proofread, Export

The walkthrough below uses the YouTube video-to-text and transcript tool and takes five steps, from pasting the link of a public video to exporting a file. Before you start, check three things:

• Account: You have to log in before you can paste a YouTube link. New accounts get 20 free minutes, with no credit card required.
• The video must be publicly accessible: Private videos, channel members-only videos, age-restricted videos, region-blocked videos, DRM-protected videos, and live streams that are still in progress can't be imported by link. Videos published less than 24 hours ago and archives of live streams that have just ended may fail to import; upload the file instead.
• Length and size: A single file can be up to 5 hours long. Check the file-size limit shown in your account’s upload dialog before uploading, and confirm that a large file can be imported before you start processing it. For costs beyond the free minutes, see Taption plans and pricing.

• Step 1 — Paste the link: After you log in, click Upload in the top-left corner. At the top of the "Upload media" dialog, switch to the Youtube icon and paste the video URL.
• Step 2 — Choose the language: In the Select the spoken language menu, pick the language spoken in the video, for example 🇺🇸 English (United States). For Mandarin videos, choose 🇹🇼 Chinese (Taiwan) when the video is mostly Mandarin with only an occasional English word, and 🇹🇼＋🇺🇸 Chinese & English (Traditional) when Mandarin and English sentences alternate frequently. The difference between the two is explained on the Mandarin and mixed Chinese–English to Traditional Chinese transcript page. The menu has no standalone Taiwanese (Hokkien) option, so videos with a lot of Taiwanese dialogue need more manual proofreading.
• Step 3 — Choose the layout: Set Transcribing method to Automated transcription, then pick a Transcript layout for your purpose. For interviews and meetings, choose Speaker transcript; you can fill in Speaker number (the total number of people speaking in the video) and Speaker label first. For a single-speaker talk or course, choose Plain transcript, which splits the text by meaning. For subtitles, choose Video subtitles. Max characters, the limit for each segment, defaults to 45 for English. Setting your own number requires a Premium or Bulk subscription, and the field isn't shown when Chinese & English is selected.
• Step 4 — Proofread: Click Create. When transcription finishes, the editor shows the video on the left and the timecoded transcript on the right. Speech recognition can still make mistakes, especially with background noise, accents, overlapping speech, or proper nouns. Check the transcript against the original audio before exporting. You can change speaker names from the Speaker menu, fix the same typo everywhere in one pass with Replace all, and highlight frequently used proper nouns and click Add to dictionary to save them to your Personal Dictionary. For the order in which to proofread a long interview, see the meeting transcript organization and efficient proofreading guide (in Traditional Chinese).
• Step 5 — Export: Click Export in the top-right corner, choose the format you need in the Export your transcript dialog, and download the file. If you are using only the free minutes, _taption is added to the exported file name, a line reading "Transcribed with Taption" is appended to the end of the file, and the timecode and speaker name toggles for TXT/PDF are unavailable. Buying minutes or subscribing removes these limits. Bilingual subtitles (Include dual subtitles) require a Premium or Bulk subscription.

This is an AI-redrawn illustration based on English-interface screenshots, not an original screenshot. Panels 1, 2, and 3 correspond to steps 1, 2, and 3 above; panel 4 corresponds to step 5. Complete the proofreading in step 4 before exporting. Create starts transcription, and Export downloads the result. The displayed value of 14 is an illustrative setting, not the default character limit for your language.

TXT, SRT, or VTT: Which Export Format Should You Choose?

Word (.docx) is not in the Export your transcript dialog. You download it from an AI Analysis result, which uses AI credits. The dialog also offers MP4, plus FCPXML and JSON, which require a Premium or Bulk subscription.

What Do TXT and SRT Files Look Like? (Format Samples)

The samples below reuse the illustrative Q&A from earlier. They only show the file formats and do not represent recognition results from a real video. Both formats hold the same content; the difference is the packaging. When a TXT file uses a custom format template (requires a Premium or Bulk subscription) that puts the timecode and speaker name on the same line, each segment is one line:

An SRT file cuts the same sentences into individual subtitle cues, each with a sequence number and start and end times (commas and periods at the ends of lines are left out in this sample):

Taption's subtitle segmentation defaults to a maximum of 45 characters per segment for English and 16 for Traditional Chinese. These are product defaults, not universal subtitle rules. For per-line character counts and punctuation conventions in Traditional Chinese subtitles, see the Traditional Chinese subtitle style guide (16 characters per line and punctuation standards). To decide whether subtitles should be delivered as a separate file or burned into the picture, see the full comparison of hard subtitles vs. soft subtitles (SRT/VTT and burned-in MP4).

If subtitles show up as garbled text in your editing software, first check whether the file displays correctly in a text editor. If the text looks right there, re-save the file as UTF-8 as your target software requires. If it is already garbled in the text editor, reopen it with the correct original encoding, confirm that the text is readable again, and only then convert it; otherwise you are just saving the garbled text in another encoding. Some tools may need UTF-8 with BOM, and the .srt and .txt files you download from Taption already start with a UTF-8 BOM. Also make sure the file wasn't saved as .srt.txt, because editing software usually won't treat that as a subtitle file.

Frequently Asked Questions

Can I get a transcript of a YouTube Short or a live stream replay?

Yes, with a few caveats. Shorts pages usually don't show the "Show transcript" button. If the Short has captions, open it in a desktop browser and change /shorts/VIDEO_ID in the URL to /watch?v=VIDEO_ID; you can then expand the description and copy the transcript just like a regular video. For Shorts or live stream replays without captions, paste the URL into Taption's Upload → Youtube dialog to run speech recognition. A live stream that is still in progress can't be imported by link, so wait until it ends. Videos that ended or were published less than 24 hours ago may fail to import; in that case, save the video as a local file first and then upload it.

My Chinese transcript has Simplified characters or Mainland China wording. How do I convert it to Taiwan Traditional Chinese?

This FAQ is for Chinese-language transcripts that need to be in Taiwan Traditional Chinese but come out with Simplified characters or Mainland China wording. The fix depends on where the text came from. For captions copied from YouTube, use a cleanup prompt like the one above, but set the output to Taiwan Traditional Chinese and have rule 4 ask for Taiwan vocabulary—for example, changing 「軟件」 and 「視頻」 to 「軟體」 and 「影片」. When you transcribe with Taption, choose one of the 🇹🇼 spoken-language options from Step 2 that output Traditional Chinese, then use Personal Dictionary and Replace all to keep specific terms consistent. If you already have a Simplified Chinese text or subtitle file, use the open-source OpenCC (Open Chinese Convert, with the s2twp Taiwan Traditional phrase conversion); its s2twp configuration converts both the characters and Taiwan-preferred vocabulary.

Can a transcript of a multi-person interview show who is speaking?

Whether speaker names appear in a YouTube transcript depends on whether the original captions include name labels. If they don't, the copied text usually won't have them either. To have a tool separate the speakers automatically, set Transcript layout to Speaker transcript when you upload to Taption. The system splits segments by voice, and once transcription is done you can change the default names to real names from the Speaker menu. The feature is described on the speaker identification page. Passages where several people talk at once or interrupt each other may still be split incorrectly and need manual fixes.

What should I do if a YouTube link fails to import into Gemini Notebook (formerly NotebookLM)?

First check whether the video qualifies. According to Google's official source import rules for Gemini Notebook (NotebookLM), it supports only public YouTube videos that have captions, and it imports only the caption text. Videos uploaded less than 72 hours ago may fail to import, and if a video is deleted or made private, the source is removed from the notebook within 30 days. If the video doesn't qualify, create a transcript with the methods in this guide and upload it as a TXT or PDF source. If the file includes timecodes and speaker names, that information becomes searchable text. The imported source is still plain text, so check any time points mentioned in the chat against the video.

Is there a copyright issue with transcribing someone else's YouTube video?

Being publicly viewable doesn't mean a video can be freely reproduced or distributed. Work first with videos you own or have permission to use. Whether any other use qualifies as fair use still depends on the purpose, the scope of the use, and applicable law; it can't be established just by calling it "personal research" or saying "the source is credited." This is general information, not legal advice.

Further Reading

• Summaries and video chapters: Once the transcript is done, you can generate a summary and chapters with AI video summaries and YouTube timestamp chapter generation, in a format like 00:00 Intro, 01:30 Topic 1, 05:10 Topic 2. According to YouTube's official Video Chapters requirements, the first timestamp must be 00:00, there must be at least three timestamps in ascending order, and each chapter must be at least 10 seconds long.
• Multilingual subtitles: To translate subtitles into other languages, see the complete guide to multilingual subtitle translation for YouTube videos (in Traditional Chinese).
• Subtitle layout: Conventions for characters per line, line breaks, and punctuation are covered in the Traditional Chinese subtitle style guide (16 characters per line and punctuation standards).

When a video has captions, Method 1 gets you the text for free, and a few minutes of cleanup takes care of the punctuation. For a video without captions, use a new account's 20 free minutes to try speech recognition and export on one of your own videos.