StaticfunctionalityAdds a microphone button to an HTMLInputElement of type 'text' or a HTMLTextareaElement for speech-to-text input via a self-hosted Whisper-Model on the Formcycle server.
Alt+A.de, en, fr). If omitted, Whisper auto-detects the language.ShowHint is true.true, shows the hotkey hint as placeholder text. Default: false.Unlike the Web Speech API variant (MEDIA.INPUT.SPEECH), this functionality processes all audio data locally on the Formcycle server (or any local server via the OpenAI /v1/audio/transcriptions API (*check the Whisper.kt Class-Documentation for more info)) using a self-hosted Whisper model via whisper.cpp.
No audio data leaves the server. There is no dependency on Google, Microsoft, OpenAI, or any other cloud provider. This makes it fully DSGVO/GDPR-compliant without requiring additional consent for cloud-based speech processing.
MEDIA.INPUT.SPEECH:| Feature | MEDIA.INPUT.SPEECH |
MEDIA.INPUT.SPEECH.WHISPER |
|---|---|---|
| Processing | Browser cloud API (real-time) | Local server (batch) |
| Data leaves device? | Yes (cloud) | No (localhost only) |
| DSGVO consent needed? | Yes (Art. 13) | No |
| Real-time interim? | Yes | Yes |
| Browser support | Chrome, Edge (limited) | All modern browsers |
Provided by the CodBi.
Provided by the CodBi. Must be an input type="text" or textarea.
Provides the Media_Input_Speech_Whisper.functionality.
Remarks
Initial Author: Callari, Salvatore (Callari@WaXCode.net) Maintainer: Callari, Salvatore (Callari@WaXCode.net)