Transcribe audio or video
Upload a file and convert speech to text — privately, in your browser.
Tip: pick the spoken language and, for songs or noisy audio, raise Accuracy to Balanced or Best — Fast (tiny) often mis-hears singing. Start with a short preview, then switch to Full file.
What to expect
- Private by design. Runs 100% in your browser — your file is never uploaded. Privacy is the upside.
- First run is slow. The first use downloads a model (~tens of MB), then it's cached, so later runs are much faster.
- Speed depends on your device. Longer files and slower computers (especially phones) take longer to process.
- Clear audio wins. Crisp speech transcribes best; background noise and heavy accents reduce accuracy.
How to transcribe a video to text
Live voice dictation
Speak and watch your words appear. Uses your browser's built-in speech recognition (works best in Chrome & Edge).
Text to speech
Type or paste text and have it read aloud using your device's built-in voices. Instant — no model download.
Downloadable audio is generated in your browser with an open-source TTS model (SpeechT5) — nothing is uploaded. The first use downloads the model once and caches it. MP3 and WAV always work; WebM/M4A/OGG appear only if your browser supports them (MP3 loads a tiny encoder the first time you pick it). The downloadable voice is English; the Play button above uses your device's built-in voices for instant multi-language preview.
How to get the most accurate transcript
A few quick do's and don'ts before you hit Transcribe.
Do
- Pick the spoken language instead of Auto — especially for non-English, mixed or accented audio. Auto can flip languages mid-file.
- Match Accuracy to the audio. Use Balanced (base) or Best (small) for music, accents or noisy clips; Fast (tiny) only for quick previews of clean speech.
- Preview first. Run the 60-second preview to check quality and speed on your device, then switch to Full file.
- Use clear, close-mic audio. Interviews, voice notes and lectures transcribe best. Convert tricky files to MP3/WAV if a video won't decode.
- Keep the tab open and awake while it runs — on phones especially, switching apps can pause processing.
Don't
- Don't expect perfect song lyrics. Speech models aren't trained on singing or music-heavy audio — they will guess and may hallucinate words.
- Don't use Fast (tiny) for important or hard audio. It's the most error-prone; only use it for quick previews of clean speech.
- Don't leave Auto-detect on for a known language — set it explicitly to avoid mixed-language output.
- Don't run very long files on a weak phone or laptop. Use a preview or split the file — it runs on your device's CPU.
- Don't treat it as word-perfect for legal, medical or captioning use without a quick human review.
Video & voice to text — your questions answered
Short, direct answers to the things people ask most.
- How do I transcribe a video to text for free?
- Open the Transcribe tab, upload your audio or video file, and click Transcribe. The tool converts speech to text directly in your browser using a Whisper model — no sign-up, no upload to a server, and no cost. When it finishes you can copy the transcript or download it as TXT, SRT or VTT.
- Is my file uploaded — is browser-based transcription private?
- Your file is never uploaded. All transcription runs locally in your browser using WebAssembly, so the audio and video never leave your device. The only network request is the one-time download of the speech model from a public CDN; your media is not part of it.
- Why is the first run slow?
- The first time you transcribe, the browser downloads the Whisper speech model — a few tens of megabytes — and caches it. That one-time download, plus the fact that transcription uses your own device's CPU, makes the first run the slowest. After that the model is cached and later transcriptions start immediately.
- What audio and video formats are supported?
- You can transcribe common formats including MP3, WAV, M4A, MP4, WebM and MOV. The tool extracts the audio track from video files automatically and resamples it to the 16 kHz mono format the model expects. Exact support depends on the audio codecs your browser can decode.
- How do I get subtitles (SRT or VTT) from a video?
- Transcribe the video, then click Download .srt or Download .vtt. The tool builds timestamped subtitle cues from the model's output, giving you a ready-to-use subtitle file you can load into YouTube, a video editor or a media player.
- Can I also dictate by voice and convert text to speech?
- Yes. The Dictate tab uses your browser's speech recognition for live voice-to-text (best in Chrome and Edge), and the Speak tab uses built-in voices to read any text aloud with adjustable rate and pitch. Both are instant and need no model download.
Need a website that brings customers?
We build websites for restaurants, hotels, gyms and local businesses. Check your current site free, or ask for a quote.
Did this tool save you transcription time?
MarketingToolsProHub keeps this tool free for everyone. If it helped you save time or get your work done, you can support its continued development.
Support is entirely voluntary. Every free tool stays free either way.