How to Transcribe Audio to Text (Free, in Minutes) — 2026 Guide

Transcribing one hour of audio by hand takes 4 to 6 hours. With a good speech recognition model, the same job takes a few minutes and you only need to review the result. This guide shows you how to do it for free, which formats you can use and how to get the best accuracy.

1. Prepare your file

Almost any recording works: MP3, WAV, M4A, OGG, OPUS, FLAC, MP4, MOV or WEBM. No conversion needed — if it's a video, we extract the audio automatically.

• WhatsApp voice message? Save it from the chat (long-press → Share → Save to Files) and upload it as is. See our guide to WhatsApp audio to text.
• Zoom or Teams meeting? Download the recording and upload the audio file (M4A) for a faster upload.
• Video for social media or YouTube? Upload it directly and download subtitles afterwards. More in video to text.

2. Upload and choose the language

Open the free tool and drop your file. The language is detected automatically, but selecting it yourself (e.g. English) slightly improves accuracy, especially in the first seconds or when there's intro music.

You can transcribe up to 10 minutes a day without signing up, and you get 30 free minutes when you create an account.

3. Review and edit

The AI gets the vast majority of words right on good-quality audio, but it's worth checking:

• Proper names and acronyms: brands, surnames or uncommon technical terms.
• Noisy passages: traffic, background music or overlapping speakers.
• Numbers: double-check important figures by listening to the clip.

In the editor, click any sentence to hear that exact moment of the audio and fix whatever you need.

4. Download in the format you need

• Word (DOCX): for reports, meeting minutes or delivering a transcribed interview.
• TXT: plain text to paste anywhere.
• SRT and VTT: timed subtitles for YouTube, Premiere, DaVinci Resolve or CapCut. Try our SRT subtitle generator.
• JSON: segment-level timestamps for developers.

Tips for a more accurate transcript

1. Record close to the source: a phone 30 cm from the speaker beats an expensive recorder across the room.
2. Avoid background noise: close windows and turn off fans or music.
3. One speaker at a time: interruptions and crosstalk cause the most errors.
4. Use decent quality: 128 kbps MP3 or higher is more than enough.

How much does it cost?

Beyond the free minutes, you can buy minute packs without a subscription: pay once and the minutes never expire. See pricing.

FAQ

What about the privacy of my recordings? Files are transferred encrypted, deleted automatically after 7 days and never used to train models.

Can I transcribe other languages? Yes, 60+ of them, such as Spanish, Portuguese or French.