Inky ClipsVideo to text, transcribed on your computer.
Open a video file from your computer and Inky Clips turns its speech into text with a model that runs in your browser. Download the words as a text transcript, an SRT or a VTT file. Free, in 99 languages, and the video is never uploaded.
It's hard to remember accurately that far back. I didn't have a lab notebook to write in when I was six, but I do remember when I was about six, my dad told me at bedtime, you know, you're made out of cells and there are chromosomes inside, and they control your future in some way, and you don't know how.
1 00:00:00,280 --> 00:00:02,120 It's hard to remember accurately that far back. 2 00:00:02,640 --> 00:00:04,390 I didn't have a lab notebook to write 3 00:00:04,410 --> 00:00:05,390 in when I was six, 4 00:00:06,060 --> 00:00:08,090 but I do remember when I was about 5 00:00:08,110 --> 00:00:09,890 six, my dad told me at bedtime,
WEBVTT 00:00:00.280 --> 00:00:02.120 It's hard to remember accurately that far back. 00:00:02.640 --> 00:00:04.390 I didn't have a lab notebook to write 00:00:04.410 --> 00:00:05.390 in when I was six, 00:00:06.060 --> 00:00:08.090 but I do remember when I was about 00:00:08.110 --> 00:00:09.890 six, my dad told me at bedtime,
How to transcribe video to text
Open your video
Sign in with Google, then choose a video from your computer. It opens in the editor without being uploaded.
Click Auto captions
The speech bubble under the preview. The first time, your browser downloads the speech model, about 1.1 to 1.6 GB, and keeps it.
Pick the spoken language
Choose from 99 languages. Naming the language gives better results than letting the model guess.
Read it through
Play the video and click any caption on the timeline to correct a word. Names and jargon are the usual fixes.
Export subtitles
Choose Text transcript, SRT or VTT. The file is made in your browser and saved straight to your computer.
Three files from one transcription
Transcribing happens through the captions in the editor, so the text you download is the text you checked. Fix a word on the timeline and every format gets the fix.
- Text transcript
- The spoken words as plain text. For show notes, articles, quotes, or pasting into a document.
- SRT
- Numbered lines with start and end times. Upload it to YouTube, or import it into Premiere Pro, DaVinci Resolve or CapCut.
- VTT
- WebVTT, the same lines in the format web players and the HTML video element read.
How accurate it is
The model is Whisper large-v3-turbo. For clear speech in English, Spanish, German, French or Polish, expect very few mistakes. Smaller languages such as Latvian, Lithuanian or Estonian need more correcting, and names, brands and jargon trip it up in any language.
It does best with one person speaking at a time and little music underneath. Every word gets its own timestamp, so subtitle lines start and end close to the speech.
Tips for smaller languagesWhat it doesn’t do
No speaker labels. The transcript doesn’t say who is talking. No translation. The text is in the language spoken. No audio-only files. The editor opens videos.
Long videos take time. There’s no length limit, but the work happens on your computer, so an hour of speech takes a while, and the tab has to stay open. A recent computer with WebGPU is considerably faster.
Want the captions on the video too?A transcript nobody else hears.
Online transcription services upload your file and run it on their servers. Here, the speech model comes to your browser instead. Interviews, meetings and private recordings stay on your computer from start to finish.
Transcribe a videoQuestions
Can I paste a YouTube or TikTok link?
No. It converts video to text from files on your own computer: a recording from your phone, a screen capture, an exported edit. Download the video first if you have the rights to it, then open the file.
Is it free to transcribe a video?
Yes. There's no per-minute price and no limit on how many videos you transcribe. You need a Google account to sign in. The transcription runs on your computer, so it costs us nothing.
Is my video uploaded?
No. The video, its audio and the transcript stay on your computer. The speech model is downloaded to your browser and runs there. Our server only handles Google sign-in.
How long can the video be?
There's no set limit. The audio is listened to in 30-second windows, so a long video simply takes longer, and the tab needs to stay open until it's done. A computer with WebGPU, such as one running a current Chrome or Edge, is much faster than one without.
How accurate is the transcript?
It uses Whisper large-v3-turbo, one of the most accurate open speech models. Clear speech in widely spoken languages comes out close to word-perfect. Background music, people talking over each other and smaller languages cause more errors, and names are often misspelled, so read it before you rely on it.
Does it label who is speaking?
No. The transcript is the words in the order they were spoken, without speaker names.
Can I transcribe an audio file or a podcast?
The editor opens video files. To transcribe a recording that is only audio, put it into a video first, for example over a still image in any video editor.
Can it translate the transcript?
No. The transcript is in the language spoken in the video.
Is this an SRT generator?
Yes. Export subtitles saves an SRT file with the timings of each line, ready for YouTube, Premiere Pro, DaVinci Resolve or CapCut. If you'd rather have the captions drawn into the video itself, export an MP4 instead.