mtsmfm.github.io

set_language

Switch the language the page transcribes speech in. Whisper is told which language to expect, so a mismatch garbles everything — call this as soon as you notice the user is speaking the other language ("ja" for Japanese, "en" for English). It takes effect on the next utterance (no restart, the model stays loaded) and is remembered in the project directory (project.json), so reopening the project keeps it. Returns {ok, language}.

Page: /ramble-to-slides/

Input schema

{
  "type": "object",
  "properties": {
    "language": {
      "type": "string",
      "enum": [
        "ja",
        "en"
      ],
      "description": "The spoken language from now on."
    }
  },
  "required": [
    "language"
  ]
}