Audio and video tools
Three of these tools turn speech into text: a transcript of a recording, a transcript of a video, or timed subtitle files. They use Whisper speech recognition on our GPU server and accept files up to 100 MB and 30 minutes. The fourth tool goes the other way and reads English text aloud.
Long files take a few minutes. You can leave the page while a transcription runs: if you come back in the same browser within 24 hours, the page reconnects to your job. Uploads and results are deleted after 24 hours, and we do not log or train on your content.
How to pick the right tool
Returns a plain transcript for MP3, WAV, M4A, OGG, FLAC and similar files. You can choose the spoken language or leave it on auto-detect.
Takes MP4, MOV, WebM or MKV and transcribes the audio track, so you do not need to extract the sound first.
Produces SRT or VTT files with timestamps, plain text, or all of them in a ZIP. Upload the SRT to most video editors and sites, and the VTT to HTML5 players.
English only, up to 5,000 characters, with American and British voices, a speed control and MP3 or WAV output.
Clear audio matters more than anything else. A close microphone, little background noise and one speaker at a time give far fewer mistakes. Always proofread names, numbers and technical terms.
Frequently asked questions
How long can a file be?
Up to 100 MB and 30 minutes per file. For longer recordings, split the file into parts and run them one after another.
Which languages are recognised?
Speech recognition covers many languages and can auto-detect them, though accuracy is highest for widely spoken ones. Text to speech supports English only.
Are the transcripts accurate?
Usually good on clear speech, weaker with noise, strong accents, overlapping voices or specialist vocabulary. Treat the result as a first draft.
What happens to my recording?
It is processed on our GPU server and deleted automatically after 24 hours. Use Delete now to remove it sooner.