Translate Text and Documents with Azure SDKs

Azure Translation SDKs for TypeScript is a pair of REST-style clients for text and document translation: TextTranslationClient and DocumentTranslationClient, each authenticating with an API key/region pair or a token credential.

The text client's /translate endpoint takes one or more input texts, an optional source language (omitted to auto-detect), and one or more target languages, and supports options like plain-vs-HTML text type, profanity handling (no action, marked, or deleted), and an LLM-specific formal/informal tone. /languages lists supported translation languages by code, name, and native name. /transliterate converts text between scripts (for example Han characters to Latin script) given a language and a from/to script pair. /detect returns the detected language and a confidence score for input text.

The document client translates a single document by posting multipart form data with a target (and optional source) language and streaming the response to a file, or runs a batch translation job across whole blob-storage containers: it generates time-limited SAS URLs for a source container (read/list) and a target container (read/write/list), submits the batch with source/target URL pairs and target languages, and reads the operation ID back from the operation-location response header. Batch status is polled by operation ID, reporting overall status and a summary of total/success counts, and the documents within a batch can be listed with pagination via the SDK's own paginate() helper. /document/formats lists supported file formats and their extensions.

Use the text client for translating, transliterating, or detecting the language of short text strings, and the document client specifically for translating whole files - either a single document inline, or many documents at scale via blob-storage batch translation with SAS-secured containers. Don't reach for the batch document flow for a single short string - that's what the text client's /translate endpoint is for.

Input is text strings (with optional source language and per-target options) for the text client, or documents/blob-container URLs for the document client. Output is translated text with a language tag, transliterated text with the target script, detected language plus confidence, or a translated document (streamed) or batch job status with per-document results. Install the client you need:

### Text translation
npm install @azure-rest/ai-translation-text @azure/identity

### Document translation
npm install @azure-rest/ai-translation-document @azure/identity

Its own best practices: omit the source language parameter to let the service auto-detect, batch multiple texts into one call for efficiency, use SAS tokens for document translation rather than public URLs, always check isUnexpected(response) before reading the response body, and use regional endpoints for lower latency.

It's the @azure-rest/ai-translation-text and @azure-rest/ai-translation-document REST clients paired with @azure/identity, and the document client integrates with @azure/storage-blob for generating SAS URLs on source/target containers.

TypeScript/Node developers building translation into an application - short-text translation and transliteration, or bulk document translation across blob storage - who want typed REST clients rather than raw HTTP calls to the Translator API.