How do I transcribe audio to text for free?

Drop your recording into the tool above, pick a Whisper model and the spoken language, then click Transcribe. The transcript appears in the box below and can be copied or downloaded as a .txt file — no account and no payment required.

Is my recording uploaded to a server?

No. The Whisper speech-recognition model is downloaded to your browser and runs there with WebAssembly or WebGPU. Your audio is decoded and transcribed locally, which you can verify in the Network tab of developer tools.

Which audio formats are supported?

Anything your browser can decode: MP3, WAV, M4A/AAC, OGG, OPUS, FLAC and WEBM. Video files with an audio track usually work too, and you can also extract the audio first with the Audio Extractor tool.

How accurate is in-browser transcription?

It uses the same Whisper models OpenAI released publicly. The tiny model is fastest and suits clear speech; the base model is slower but noticeably better on accents, background noise and technical vocabulary. Always proofread the result before relying on it.

Why is the first run slow?

The model weights (roughly 40–80 MB depending on the model) download once and are then cached by the browser. Later transcriptions start immediately and work offline.

Audio to Text — Free Private Transcription

Transcription without the upload

Most transcription services ask you to upload the recording, wait in a queue and hand over a card for anything past a few free minutes. Morphix takes the opposite route: the Whisper model is downloaded into your browser once, then your audio is decoded to a 16 kHz mono stream and transcribed locally. Sensitive material — client calls, medical notes, interviews under embargo — never touches a server.

Choosing a model

Fast (tiny) — best for clear, single-speaker recordings and quick drafts. Roughly real-time on a modern laptop.
Accurate (base) — better with accents, crosstalk and jargon, at roughly twice the processing time.
Set the spoken language explicitly. Auto-detection on short clips is the most common cause of garbled output.
Very long files are memory-hungry. For recordings over ~30 minutes, split them first and transcribe in parts.

Transcribing video

Video files with an audio track usually load directly. If a container gives your browser trouble, pull the soundtrack out first with the Audio Extractor and drop the resulting MP3 or WAV here.

Audio to Text Transcription

Transcription without the upload

Choosing a model

Transcribing video

Related tools