01 / BEFORE YOU BEGIN
Start with a captioned video.
WatchToText reads captions already available on a YouTube video. You will need a video link or its 11-character ID. You do not need an account.
This is caption extraction, not fresh audio transcription. If a video has no usable caption track, there may be no transcript to retrieve.
Automatic captions can mishear names, technical terms and punctuation. Check anything you plan to quote against the original video.
02 / GET THE WORDS
One link. A few simple steps.
Copy the YouTube link.
Use the full watch link, a shortened youtu.be link, or the video ID itself.
youtu.be/VIDEO_ID
Illustrative URL patterns — replace VIDEO_ID with the real ID.
Paste it into WatchToText.
Open the homepage, select YouTube and paste your link into the video field. Choose Get the text.
Read the result.
When captions are available, the transcript appears below the input. If it cannot be retrieved, read the status message to see what happened.
Take the words with you.
Copy the text, download a TXT file, or save the timed captions as SRT. Choose the format for what you want to do next.
03 / CHOOSE HOW TO SAVE
Words for reading.
Timing when you need it.
Copy: a quick handoff
Put the transcript on your clipboard and paste it into a note or document. Copying does not save a file to your device.
TXT: your reading copy
A plain-text file keeps the caption text in order without timestamps. Open it in a text editor or bring it into your notes.
SRT: the timed version
A SubRip file includes numbered captions and their start and end times. Timing comes from the caption track, and may need editing for your subtitle workflow.
04 / IF IT DOES NOT WORK
Read the message.
Then choose the next step.
The link is not recognized
Check that you pasted a YouTube watch link, a youtu.be link, or the 11-character video ID. Remove any surrounding text and try again.
No usable captions, or an empty track
Check whether the video has captions available on YouTube. If the track is missing or empty, try another captioned video. A player caption control alone does not guarantee this tool can retrieve a track.
The video cannot be played here
Try a public watch link that opens for you on YouTube. Private or otherwise inaccessible videos may not return a usable caption track.
YouTube refused the request
Wait before trying again. Availability can change, and retrying is not a guarantee that the request will succeed.
FIELD TEST · SEPTEMBER 19, 2026 UTC
What if there is no caption track?
The main route is to retrieve an existing platform caption track. If no usable track exists, speech recognition can create text from audio you have permission to process. A failed caption request alone does not prove that a track is missing.
This is a separate local test, not a feature of our YouTube tool. WatchToText's YouTube tool still retrieves existing captions only. This test does not verify YouTube caption discovery or add an audio-transcription fallback to the website.
The source and the missing track
We used Mark F. Smith's human reading of Abraham Lincoln's The Gettysburg Address (version 3), from this Internet Archive item. LibriVox labels the recording public domain in the USA. It is a volunteer reading, not Lincoln's recorded voice.
At 20:11 UTC on September 19, 2026, the item's successfully retrieved metadata listed 31 files, none with a timed-subtitle extension or caption format. Its player playlist also listed no subtitle tracks. Untimed speech text and cover OCR exist. This finding applies to that item at that time, not every copy of the recording.
One local run: result and cost
We took source time 00:00:30 to 00:01:00 from the 161.241625-second recording, without editing sentences. The 30-second input was converted to 16 kHz mono PCM WAV and processed on an Apple M2 Mac with whisper.cpp 1.8.4 and the non-quantized large-v3-turbo model. The process exited with code 0.
Measured process wall time was 9.721103042 seconds, including startup, model loading and output. Recognition API calls: 0; recognition API cost: 0. Electricity and other local costs were not measured. These figures come from this single run; they are not a long-recording performance estimate or an accuracy benchmark.
Complete raw output for the input clip
The timestamps are relative to the 30-second input; add 30 seconds to locate them in the source. Text and timing below are unchanged, including the mid-introduction opening. No proofreading was applied.
1 00:00:00,000 --> 00:00:10,420 address by Abraham Lincoln. Fourscore and seven years ago our fathers brought forth upon this 2 00:00:10,420 --> 00:00:17,700 continent a new nation conceived in liberty and dedicated to the proposition that all men are 3 00:00:17,700 --> 00:00:26,120 created equal. Now we are engaged in a great civil war, testing whether that nation or any 4 00:00:26,120 --> 00:00:30,000 nation so conceived and so dedicated can long endure.
The local setup can produce timestamped text from this public audio sample. Chinese accuracy, long recordings and a no-caption YouTube video were not tested here. Check generated words and timing against the audio before quoting or publishing them.
05 / A FEW QUESTIONS
Before you go.
Can I use just the video ID?
Yes. Paste the 11-character YouTube ID into the same field you use for a full link.
Will this write a transcript from the audio?
No. The YouTube tool retrieves an existing caption track. It does not create one for a video without usable captions. See our separate local speech-recognition test for a fallback example outside the website.
Do I need to sign in?
No account is needed to use WatchToText or save a transcript.
Can I follow these steps on a phone?
The basic flow is the same: paste a link, request the text, then copy or save a successful result. Your browser and device determine how saved files are opened.
READY WHEN YOU ARE
Keep the part that stays with you.
A useful sentence. A new idea. A moment you want to come back to.
Get the text ↗