Output language

By default, summarize writes the summary in the same language as the source content (--language auto). If language detection is uncertain, it falls back to English.

This affects the language of the generated summary text (not extraction/transcription).

CLI

summarize <input> --language auto
summarize <input> --language en
summarize <input> --language de
summarize <input> --language english
summarize <input> --lang german

Supported inputs (best-effort):

• auto (default): match the source language
• Common shorthands: en, de, es, fr, …
• Common names: english, german/deutsch, spanish, …
• BCP-47-ish tags: en-US, pt-BR, …
• Free-form hints: German, formal

Config default

Preferred:

{
  "output": { "language": "auto" }
}

Legacy (still supported):

{
  "language": "en"
}

Unknown values are passed through to the model (sanitized).