Implement text analysis solutions 4.1 Apply language model text analysis
A legal firm must translate hundreds of Word and PDF files while keeping each document's…
Study guide skill: Build solutions that translate text by using Azure Translator in Foundry Tools or LLM-powered translation flows
Question B1-039
A legal firm must translate hundreds of Word and PDF files while keeping each document's original structure and formatting. What should you use?
Select one answer.
Answer: C
Document translation translates batches of complex files while preserving the structure and format of the original documents, and the batch process requires an Azure Blob Storage account with containers for source and translated documents. Text translation works on strings, so formatting would be lost.
Deep dive
How it works
Document translation is the Azure Translator feature for files. It translates multiple, complex documents while preserving the original document structure and data format. It works in two modes:
- Asynchronous batch translation handles many documents and large files. It needs an Azure Blob Storage account with containers for the source and translated documents, and you authorize access to the storage URLs with SAS tokens or managed identities.
- Synchronous single-file translation handles one file at a time, optionally with a glossary. It doesn't need Blob Storage, and the translated document comes back directly in the response.
For hundreds of Word and PDF files, use asynchronous batch translation. It can also translate text inside images in .docx and .pptx files, and it can apply custom translation models.
The other options, and when you'd use them
- Synchronous text translation, sending each file's text: Text translation translates between supported languages in real time, and it's the right choice for strings in an app, with features such as a dynamic dictionary. Sending extracted text gives up the layout that document translation preserves.
- Custom Translator with a dictionary only: Custom Translator builds customized translation systems for domain- and industry-specific terminology, and you can train a model on dictionary data alone if you have no other training data. Models are accessed through the Translator Text API v3. It improves what the translation says, but it doesn't handle files. You can pair it with document translation.
- Azure Speech translation: Speech translation does real-time, multi-language speech-to-speech and speech-to-text translation of audio streams. Use it for spoken audio, not documents.
Exam tip
"Many files, keep the formatting" means asynchronous document translation with source and target Blob containers. "One file, returned in the response" means synchronous document translation, and "domain terminology" means Custom Translator.
References
Last reviewed: