Speech to Text
Private live dictation when your browser supports on-device speech recognition.
- Runs in your browser — files never leave your device
- Fewer clicks, keyboard friendly
Start or stop when focus is outside a text field: Space
Private mode: this tool only starts recognition when the browser confirms on-device processing. Audio and text are not sent to a recognition service.
How to use
- Choose the language you will speak.
- Select Start dictation and allow microphone access if your browser asks.
- Speak naturally; finalized phrases appear in the editable transcript.
- Select Stop dictation, then edit, copy, or download the text as a .txt file.
Why this tool requires on-device recognition
The usual Web Speech API setup can send microphone audio to a recognition service. This tool deliberately does not use that mode. Before recording, it asks the browser for an on-device dictation language pack and sets SpeechRecognition.processLocally to true. If the pack or API is unavailable, recording stays disabled instead of silently switching to cloud recognition. The first setup can download a language pack; after it is installed, recognition can work without sending audio or transcript text to a recognition service.
Browser support and language packs
SpeechRecognition is not a Baseline web feature, so support differs by browser and device. The on-device version also needs a language pack for the selected language. The Web Speech API exposes available() to check a pack and install() to request it; this tool uses both before it starts. A browser that has only the older, cloud-backed or prefixed recognition API is intentionally shown as unsupported because this page keeps dictation private.
Getting a cleaner transcript
Speak one thought at a time, keep the microphone close, and reduce music or overlapping voices. Dictation is best for notes, drafts, and short messages; review names, numbers, technical terms, and punctuation before sharing. The text area stays editable while you work. Your transcript is saved in this browser’s local storage for refresh recovery, and Clear removes that saved copy. Copy or download the final text when you are ready to use it elsewhere.
Frequently asked questions
Does this speech-to-text tool upload my voice?
No. This page starts speech recognition only after the browser reports that it can process the selected language on-device. If it cannot, the tool refuses to record instead of using a cloud recognition service.
Why does the tool say private dictation is unavailable?
On-device Web Speech API recognition has limited browser support and requires a compatible language pack. The tool needs both the available() and install() browser features so it can guarantee local processing rather than sending microphone audio to a service.
Is my transcript saved after I refresh?
Yes. The editable transcript is stored only in this browser’s local storage after you stop or type. It is not uploaded, and selecting Clear removes the saved transcript from this browser.
Can I transcribe a recording or an audio file?
No. This tool is for private live dictation from your microphone. File transcription needs a speech-recognition model and can involve large local downloads, so it is intentionally kept out of this focused tool.
Why should I proofread dictated text?
Speech recognition can mishear names, numbers, accents, punctuation, and words spoken over background noise. Review the editable transcript before using it for an application, legal document, medical note, or other important record.
Related tools
Updated 2026-10-27