LaraDocumentTranslator

Translate Haystack Documents using Lara, an adaptive translation API that combines LLM fluency with low hallucination and latency.

Key Features

Translates Document text content using the Lara translation API.
Auto-detects source language when no source language is specified.
Supports translation memories and glossaries for consistent domain-specific terminology.
Offers three translation styles: faithful (accuracy), fluid (readability), and creative (expressive).
Accepts per-document translation settings via list parameters for batches with mixed language or style requirements.
Preserves original Document metadata and adds original_document_id to translated Documents.

Configuration

Sign up at laratranslate.com and get an access key pair.
Set the LARA_ACCESS_KEY_ID and LARA_ACCESS_KEY_SECRET environment variables. For instructions, see Create Secrets.
Configure LaraDocumentTranslator in your pipeline YAML with the source and target language codes.

Connections

LaraDocumentTranslator takes a list of Document objects as documents input and outputs a list of translated Document objects as documents.

Connect a document converter's or retriever's documents output to the documents input. Connect the documents output to downstream processing components such as a document writer or embedder.

Source Code

To check this component's source code, open document_translator.py in the Haystack Core Integrations repository.

Usage Examples

Basic Configuration

  lara_translator:
    type: haystack_integrations.components.translators.lara.document_translator.LaraDocumentTranslator
    init_parameters:
      access_key_id:
        type: env_var
        env_vars:
        - LARA_ACCESS_KEY_ID
        strict: false
      access_key_secret:
        type: env_var
        env_vars:
        - LARA_ACCESS_KEY_SECRET
        strict: false
      source_lang: en-US
      target_lang: de-DE
      style: faithful

Using the Component in a Pipeline

# haystack-pipeline
components:
  pdf_converter:
    type: haystack.components.converters.pypdf.PyPDFToDocument
    init_parameters: {}
  lara_translator:
    type: haystack_integrations.components.translators.lara.document_translator.LaraDocumentTranslator
    init_parameters:
      access_key_id:
        type: env_var
        env_vars:
        - LARA_ACCESS_KEY_ID
        strict: false
      access_key_secret:
        type: env_var
        env_vars:
        - LARA_ACCESS_KEY_SECRET
        strict: false
      source_lang: en-US
      target_lang: de-DE
  writer:
    type: haystack.components.writers.document_writer.DocumentWriter
    init_parameters:
      document_store:
        type: haystack_integrations.document_stores.opensearch.document_store.OpenSearchDocumentStore
        init_parameters:
          hosts:
          - ${OPENSEARCH_HOST}
          index: translated-docs
          embedding_dim: 768
          create_index: true

connections:
- sender: pdf_converter.documents
  receiver: lara_translator.documents
- sender: lara_translator.documents
  receiver: writer.documents

inputs:
  files:
  - pdf_converter.sources

max_runs_per_component: 100

metadata: {}

Parameters

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
`access_key_id`	Secret	`Secret.from_env_var("LARA_ACCESS_KEY_ID")`	Lara API access key ID.
`access_key_secret`	Secret	`Secret.from_env_var("LARA_ACCESS_KEY_SECRET")`	Lara API access key secret.
`source_lang`	str \| None	`None`	Source language locale code (for example, `en-US`). If `None`, Lara auto-detects the language. See the supported languages list.
`target_lang`	str \| None	`None`	Target language locale code (for example, `de-DE`). See the supported languages list.
`context`	str \| None	`None`	Optional context text sent to Lara to improve translation quality. This text is not translated.
`instructions`	str \| None	`None`	Optional natural-language instructions to guide translation style or terminology (for example, "Be formal").
`style`	str	`faithful`	Translation style. One of `faithful` (accuracy and precision), `fluid` (natural readability), or `creative` (artistic expression).
`adapt_to`	list[str] \| None	`None`	Optional list of translation memory IDs for domain adaptation at inference time.
`glossaries`	list[str] \| None	`None`	Optional list of glossary IDs for consistent terminology enforcement across translations.
`reasoning`	bool	`False`	If `True`, uses the Lara Think model for multi-step linguistic analysis. Increases latency and cost.

Was this page helpful?

Key Features​

Configuration​

Connections​

Source Code​

Usage Examples​

Basic Configuration​

Using the Component in a Pipeline​

Parameters​

Init Parameters​

Related Information​