How to transcribe audio to text
Every platform has a way to turn speech into text now, at very different prices. Here is what each one offers, and a free option that never uploads the recording.
What you are choosing between
An interview, a lecture, a voice memo or a meeting recording all turn into text the same way underneath: a speech-recognition model listens and writes down what it hears. Where that model runs, and who pays for it, is what actually differs between the options below: on the device for free, in the cloud for a subscription, or by a person for a fee.
On an iPhone
The Voice Memos app transcribes its own recordings automatically on-device (iOS 17 and later): open the recording and tap the transcript icon. It only works for audio recorded in Voice Memos itself, not a file someone sent you, and there is no SRT or VTT export for video captions. Live Speech and Dictation cover live transcription while you talk, not an existing file.
On Android
Google's Recorder app (on Pixel and some other phones) transcribes as you record and lets you search the text afterward, on-device. Outside Recorder, Android has no system-wide "transcribe this file" option; Google Docs voice typing only works live through the microphone, not on an existing recording.
On Windows or a Mac
Word's Transcribe feature (Microsoft 365, in Word on the web) uploads the file to Microsoft and returns text with speaker labels, with a monthly transcription-minutes cap on most plans. Apps like Otter.ai and Rev do the same over the web, priced by the minute or by a subscription tier; Rev's human transcription is more accurate for difficult audio but costs several dollars per minute.
Transcribing without uploading or paying per minute
Stayput's transcribe tool runs Whisper, OpenAI's speech-recognition model, directly in the browser tab: drop an MP3, WAV, M4A, MP4 or MOV file, and the model downloads once and then works entirely on your device. It writes plain text (see audio to text and video to text), or timed SRT subtitles and WebVTT captions for a video, in over a dozen languages, with no per-minute cost and no upload of the recording itself.
It does not label who is speaking, and it struggles with the same things every automatic transcriber does: heavy accents, overlapping speech, and specialist terms. For a recording where the wording actually matters, read the result against the audio before using it.
Questions
Can I transcribe audio for free without a time limit?
Yes, with a tool that runs the speech model in the browser rather than metering server time; Stayput's transcribe tool has no per-minute cost because nothing is processed on a server.
How accurate is automatic transcription?
Clear speech in a widely spoken language comes out well. Strong accents, background noise, crosstalk and technical vocabulary reduce accuracy with any automatic transcriber, free or paid.
Does transcribing a recording upload it anywhere?
With cloud services like Word's Transcribe or Otter.ai, yes, the audio is sent to their servers. A browser-based tool that runs the model on your device, like Stayput's, never uploads the file.
Can I get subtitles, not just plain text?
Yes. Stayput's transcribe tool can output SRT subtitles or WebVTT captions timed to the speech, ready to load into a video editor or a web page.