How to turn a voice message into text — privately

A two-minute voice note is quick to send and slow to receive — especially in a meeting, on a train, or when you need to find one detail later. A transcript can be skimmed, searched and copied. The catch is that most transcription services upload your audio to a server, and voice messages are often private.

Transcription on your own device

The Voice to Text Converter runs a speech recognition model called Whisper inside your browser. The model is downloaded once — about 40 MB for the faster version and about 80 MB for the more accurate one — and then cached. Your audio is decoded and transcribed on your own processor, and it is not sent to this site or anyone else.

Step 1: save the voice message as a file

• WhatsApp on a phone: press and hold the voice message, choose Share, and save it to your files.
• WhatsApp on a computer: open the message menu and choose Download.
• Other apps generally have a similar share or save option for voice messages.
• Phone recordings from a voice memo app can be shared or exported as a file.

MP3, M4A and AAC, WAV, OGG and Opus voice notes, WebM and FLAC all work — anything your browser can decode. If a file will not open in one browser, try another or convert it to MP3.

Step 2: transcribe it

1. Add the audio file.
2. Choose the language, or leave it on automatic detection.
3. Pick the faster model for quick notes, or the more accurate one for names and detail.
4. Press Transcribe. The first time, the model downloads; after that it starts straight away.
5. Copy the text, or download it — with timestamps and subtitles if you turned them on.

A one-minute clip takes a few seconds on a recent laptop and longer on an older phone. Recordings up to 10 minutes are accepted.

Getting an accurate transcript

• Clear audio matters most. One person speaking close to the phone, with little background noise, transcribes far better than a recording across a busy room.
• Choose the language when you know it. Automatic detection can be misled by a short clip or a greeting in another language.
• Use the more accurate model for names, numbers, addresses and technical terms.
• Split long recordings into shorter clips if you have more than ten minutes.

Check before you rely on it

Speech recognition is very good on clear speech and still makes mistakes. It can mishear names and specialist words, struggle when people talk over each other, and occasionally invent words during long silences. Read the transcript before quoting it, and check every name, number and date against the audio.

Tidying the text

Paste the transcript into Text Cleaner to remove stray spaces or join broken lines, and into the Word Counter if you need its length or reading time.

Built-in transcription versus a file

Some messaging apps and phones now offer transcripts of voice messages themselves, for certain languages. If yours does, it is the quickest route for a single message. A separate tool is useful when the built-in option is not available in your language or on your device, when you are working on a computer, when the recording is a file rather than a message, or when you want timestamps and subtitles.

Making use of the transcript

• Search it for the one detail you needed — an address, a time, a number.
• Quote it accurately, using the timestamps to find the exact moment in the audio.
• Make subtitles for your own short videos, then check the timing and spelling before publishing.
• Keep a record of spoken instructions or agreements, alongside the original audio.

Respect the speaker

A voice message was sent to you, not to be published. Transcribing it for your own convenience is one thing; sharing the transcript is another. Ask before passing on what someone said, and check the rules that apply to you before transcribing recordings of other people.