Transcribe

Transcribe is the ultimate 'do one thing well' tool: it's literally a single HTML page that sends your audio to the Whisper API and returns text. We'd reach for this when we want to quickly transcribe a meeting recording or test speech-to-text without setting up a server or signing up for another SaaS monthly plan. You provide the API key, it saves to localStorage, and you're done. The language support is generous—over 90 languages listed—and you can switch between automatic detection (slower) or manual selection. Output in SRT or VTT is handy for subtitles. Where it bites: there's no diarization, no editing timeline, no file management. If you transcribe a lot, you'll notice the lack of batch processing—you upload one file at a time. Also, OpenAI's documentation explicitly warns against exposing API keys in client-side code, so security-minded users should be aware this is for personal, low-risk use only. Compared to something like MacWhisper or WhisperX, Transcribe is lighter but less capable offline—it depends entirely on the API. Best for quick one-offs and developers who want to peek at the code. Skip it if you need high-volume transcription or don't want to handle your own API billing.