WhatsApp voice notes to text, on your computer

How it works

Click Transcribe under a voice note. By default the speech model runs on your computer and the audio stays there. The first time, WA Folders shows the download size and asks before it fetches the model once from Hugging Face: Accurate (about 252 MB) or Light (about 80 MB, better for slower computers or metered connections).

Prefer a cloud model? Add your own OpenAI or Google Gemini key. Then only the voice note you click goes from your browser to that provider.

Where transcripts go

Transcripts are saved on this device. They show up in WA Folders search and in exports. Voice notes longer than 15 minutes aren't transcribed.

Free and Pro

Free: 5 transcriptions a day. Pro: unlimited, for a one-time $4.99.

Set it up in two minutes

1. Install WA Folders and open web.whatsapp.com.
2. Open a chat with a voice note and click Transcribe under it.
3. The first time, choose Accurate or Light and confirm the one-time download. WA Folders shows the size before it fetches anything.
4. Read the text under the voice note. Click Translate under it if it's in another language.
5. Search for words from it later in the WA Folders panel.

When it helps

• You're in a meeting or on a train without headphones and a two-minute voice note arrives.
• Someone sent an address or a list in a voice note and you need to copy it.
• You want to find what was said in a voice note last week. Transcripts are searchable.
• The voice note is in Portuguese or Spanish: transcribe it, then click Translate under the transcript.

Accurate or Light

The first time you click Transcribe, WA Folders asks which on-device model to download. Accurate (about 252 MB) handles accents and noisy audio better. Light (about 80 MB) is faster on older laptops and better on metered connections. You download it once; after that, transcription works offline on that computer.

Speed depends on your computer. A short note usually comes back in seconds; long ones take longer. Notes over 15 minutes are skipped.

Your own API key (optional)

If your computer is slow, add an OpenAI or Google Gemini key in Settings. Then the audio of the one voice note you click goes from your browser straight to that provider, which bills you directly. WA Folders never sees it and never sends anything without a click.

Privacy

With the default on-device model, the audio never leaves your computer. Transcripts are saved on this device, appear under the voice note, and are included in search and exports. They don't sync to your phone.

Limits and honest caveats

Transcripts aren't perfect. Names, numbers and mixed languages are where mistakes show up, so check anything you'll act on. Free includes 5 transcriptions a day; Pro is unlimited for a one-time $4.99.

Related: translate WhatsApp Web messages, search messages and transcripts, and the WhatsApp blog.

WhatsApp's own transcripts

WhatsApp's phone apps can show voice-message transcripts for some languages and devices, and the feature is off by default. WA Folders adds transcription on the computer, and keeps the text searchable alongside your chats.

Free vs Pro for voice notes

Common problems

• The download is slow: pick Light (about 80 MB) instead of Accurate.
• The text has mistakes: clear audio helps most; Accurate does better than Light on accents and background noise.
• No Transcribe button: it appears under voice notes in an open chat on web.whatsapp.com, not on the phone app.

Questions