How To
How to Transcribe a Video for Free
Free video transcription powered by Whisper AI — runs locally in your browser with complete privacy.
Step-by-Step Instructions
- 1
Open VideoCaptions.AI and create a new project
Go to videocaptions.ai and click create new project. The import wizard opens automatically, ready for your file. No account creation or signup is required.
- 2
Upload your video file
Drag and drop your video into the upload area. The tool extracts audio from the video using WebCodecs — this happens locally in your browser. Your video file never leaves your device.
Tip: For faster extraction, shorter clips process quicker. If you only need a portion transcribed, trim the video before uploading.
- 3
Select a Whisper model and transcribe
Choose tiny for fastest speed, base for the best balance, or small for highest accuracy. The model downloads once and is cached for future use. Click transcribe and Whisper processes your audio locally.
- 4
Review your transcript
The transcript appears with word-level timestamps. Each word is individually editable — fix any errors, remove filler words, and adjust timing. The visual editor shows the transcript synced with your video playback.
- 5
Use your transcript
Continue in VideoCaptions.AI to style and export captioned video, or copy the transcript text for use in blog posts, show notes, or other written content. The word-level timing data is preserved throughout the workflow.
01
Why Transcribe Videos with Whisper AI?
Video transcription has historically been expensive and time-consuming. Professional transcription services charge per minute of audio, and even budget online tools require uploading your content to cloud servers where it is processed and stored by a third party. Whisper AI changes this equation by providing state-of-the-art transcription accuracy as a free, open-source model that can run locally. VideoCaptions.AI takes this further by running Whisper directly in your web browser using WebAssembly — no installation, no server upload, and no cost. Your video stays on your device throughout the entire process. The quality is remarkable for a free tool. Whisper was trained on 680,000 hours of multilingual audio data, giving it robust accuracy across 99 languages, diverse accents, and varying audio conditions. For most clear speech content, the transcription is accurate enough to use with only minor corrections, saving hours compared to manual transcription.
02
Getting the Best Transcription Results
Transcription accuracy depends on several factors that you can optimize. Audio quality is the most important — clear recordings with minimal background noise produce the best results. If your video has music or ambient sound competing with speech, accuracy will decrease. Using an external microphone rather than a built-in laptop mic makes a significant difference. Speaking pace also matters. Normal conversational speed transcribes well, but very rapid speech or mumbling can cause errors. The Whisper model you select affects both speed and accuracy. The tiny model is fastest but makes more mistakes. The base model is the recommended default for English content, offering an excellent speed-to-accuracy ratio. The small model is best for non-English languages, accented speech, or content where maximum accuracy justifies the extra processing time. After transcription, always review the output. Even the best AI makes occasional errors with proper nouns, technical terms, and homophones. A quick review pass catches these issues and ensures your transcript is publication-ready.
Frequently Asked Questions
Everything you need to know before you start.
Can't find what you're looking for? Contact us
Yes. Whisper AI runs locally in your browser at zero cost. There are no per-minute charges, no subscription fees, and no limits on how many videos you can transcribe. The tool is free because all processing happens on your device — there are no server costs to pass on.
No. Everything runs in your browser. Audio extraction and Whisper transcription happen locally on your device. Your video file never leaves your computer. This provides complete privacy — no one else has access to your content at any point in the process.
Whisper supports 99 languages including English, Spanish, French, German, Hindi, Arabic, Japanese, Korean, Mandarin, Portuguese, and many more. English has the highest accuracy. Use the small model for the best results with non-English languages.
Speed depends on your device and the selected model. The base model transcribes one minute of audio in roughly 15 to 30 seconds on a modern laptop. Longer videos take proportionally more time. The model only needs to download once and is cached in your browser for future use.