Language Identification

Two pipelines cover language identification — one for text-based files and one for audio. Both return the same output schema with detected languages, confidence scores, and a primary language.

lt-language-id-text

Detects the language(s) present in text and document files using an LLM. Handles multilingual documents and returns a ranked list of detected languages.

100 credits per file · ~1 minute

Accepted Files

Type	MIME types
Text	`text/*`
Documents	`application/pdf`, Word (`.docx`)

Up to 10 files per run.

Context Parameters

None.

API Example

POST /v2/lt-pipelines/project/{projectId}/run

{
  "pipelineSlug": "lt-language-id-text",
  "fileIds": ["file_abc123"],
  "contextData": {}
}

Output

Both pipelines return the same output schema:

{
  "pipelineSlug": "lt-language-id-text",
  "output": {
    "detectedLanguages": [
      {
        "language": "English",
        "code": "en",
        "confidence": 0.95
      },
      {
        "language": "French",
        "code": "fr",
        "confidence": 0.42
      }
    ],
    "primaryLanguage": "English",
    "primaryLanguageCode": "en",
    "multilingual": true
  },
  "rawText": "English"
}

Output Fields

Field	Type	Description
`detectedLanguages`	array	Ranked list of detected languages with confidence scores
`detectedLanguages[].language`	string	Language name (e.g. `"English"`)
`detectedLanguages[].code`	string	ISO 639-1 code (e.g. `"en"`)
`detectedLanguages[].confidence`	number	Confidence score 0–1
`primaryLanguage`	string	Name of the most confident detected language
`primaryLanguageCode`	string	ISO 639-1 code of the primary language
`multilingual`	boolean	`true` if more than one language was detected
`rawText`	string	Primary language name as plain text