How to convert audio to text

The quickest way to convert audio to text is to upload the recording to an AI transcription service: you get a transcript with timestamps back in minutes. If you only have a short clip, the dictation tools in Microsoft Word or Google Docs can do it too, and for word-perfect legal or medical records you can still transcribe by hand. Here is how each method works and when to use it.

Published October 29, 2026 · Stenovo

Method 1: Upload it to an AI transcription service

This is the right choice for interviews, lectures, meetings, podcasts and voice notes: anything longer than a few minutes. Here is how it works in Stenovo:

1. Open Stenovo. It runs in your browser, so there is nothing to install.
2. Upload the audio file. MP3, M4A, WAV, AAC, OGG, OPUS and WMA all work, and so do video files. You can also press Record and speak, or use the browser extension to capture audio playing in a tab, such as a webinar.
3. Choose the language, or leave it on automatic. Stenovo detects the language on its own; choosing it yourself helps accuracy.
4. Wait a few minutes. The transcript appears with timestamps, along with a summary of the key points.
5. Download or copy the text. Save it as TXT or PDF, with or without timestamps, or as SRT subtitles.

On clear audio this is about 98% accurate, on par with a professional human transcriber. Convert your audio to text with Stenovo.

Method 2: Use Microsoft Word or Google Docs

If you already work in one of these editors, you can turn short recordings into text without another tool.

Microsoft Word

In Word for the web, Microsoft 365 subscribers can open Home, then Dictate, then Transcribe, and upload an audio file. Word returns the text split by speaker, which you can then edit and add to the document.

Google Docs

Google Docs has voice typing under Tools, but it listens to your microphone rather than reading a file. To transcribe a recording, you play it out loud next to the mic, in real time, and accuracy drops because the audio is recorded a second time. It works for a quick clip, not for an hour-long interview.

Method 3: Transcribe it by hand

Typing it out yourself still makes sense when every word must be exact and verified, as in legal or medical records, or when the audio is too poor for software. Expect to spend several hours on each hour of audio.

A common middle ground is to run the file through an AI service first and then correct the draft by hand.

Which method should you use?

Tips for a more accurate transcript

• Record close to the microphone. Distance and echo cost more accuracy than an inexpensive mic does.
• Cut background noise. Music, traffic and air conditioning all compete with the voice.
• Let one person speak at a time. Overlapping voices are the hardest part for any transcriber, human or AI.
• Check names and technical terms. They are where even a good transcript most often slips.

Common questions