Transcribe Video and Audio to Text

Kai The Scribe transcribes a video or audio file into plain text in about a minute: sign in with Google, paste a public media URL or upload a file up to 4 MB, click Start Transcription, then copy the transcript or download it as .txt.

The transcription tool is the one live feature on this site. The transcription tool runs on Deepgram speech recognition with punctuation and number formatting turned on. Every account gets 60 free minutes a month, and Pro is $19 a month for more.

Open the transcription tool

What you need before you start

• A Google account for sign-in.
• A recording as MP3, WAV, M4A, MP4, MOV, or WEBM: under 4 MB to upload, or any size at a direct public link.
• Or a direct public link to that file, for example https://example.com/episode.mp3.

How to transcribe a video or audio file

There are six steps. Each step names the exact label you will see on the page.

1. Open the tool and sign in with Google. Go to the transcription tool and click Sign in with Google. Sign-in keeps your free minutes and transcripts tied to one account.
2. Upload a file or paste a media URL. Choose a file under Upload File (MP3, WAV, M4A, MP4, MOV, or WEBM, up to 4 MB), or paste a direct public link to the file in the Media URL field. Use one or the other, not both.
3. Pick a language. Leave Language on Auto Detect for most recordings. Pick the language yourself if the recording is short or mixes languages.
4. Turn on speaker labels for interviews. Tick Label different speakers under Speaker Labels if the recording has more than one voice. Leave it off for a solo recording.
5. Click Start Transcription. The button reads Uploading and transcribing while the file is sent and processed. Most recordings finish within a minute.
6. Copy or download the transcript. Read the text under Your Transcript. Click Copy to put it on the clipboard, or click Download .txt to save a plain text file named after the source.

Supported inputs

What you get back

The result has two parts. The first part is a summary card with the recording duration, the detected language, and the source file name or URL. The second part is the transcript itself as one block of punctuated plain text. Two buttons sit above it: Copy puts the text on your clipboard, and Download .txt saves it as a plain text file named after the source. Example: uploading interview-ep12.mp4 downloads as interview-ep12.txt.

Limits

• 4 MB per upload. That is about four minutes of a 128 kbps MP3. For anything longer, host the file and paste its URL under Media URL.
• 60 seconds of processing per request. Most recordings under an hour finish well inside that window. Split a very long file into parts if a request times out.
• Direct file links only. The URL field fetches the file itself. A share page, a playlist, or a streaming link returns an error.
• Plain text output. The transcript is text, not an SRT or VTT caption file. With speaker labels on, each turn starts with Speaker 1:, Speaker 2: and so on, numbered rather than named.
• Copy before you leave. If the page says the transcript is not saved to your account yet, copy or download it before closing the tab.

Pricing

Free accounts get 60 minutes of transcription a month. Pro is $19 a month for creators who publish weekly. Minutes count the length of the recording. Full details are in the pricing section.

Tips for a cleaner transcript

1. Record each speaker on a separate microphone when you can. Crosstalk is the biggest source of errors.
2. Remove the music bed before uploading. Speech under music transcribes worse than dry speech.
3. Set the language yourself for recordings under two minutes. Auto Detect has less audio to work with.
4. Search the transcript for names and product terms first. Those are the words most often misheard.

Frequently asked questions