Amazon Translate Guide: Languages, Features and Pricing

Amazon Translate covers 75 languages in real time or in batch. Learn its features, pricing, use cases and best practices.

What is Amazon Translate?

Reaching global audiences means speaking their language, and product descriptions, support conversations, legal documents, user interfaces and marketing content all need localizing. That used to mean expensive human translation agencies or custom machine translation pipelines. Amazon Translate, however, removes both obstacles with neural machine translation that works at scale.

Amazon Translate is a fully managed neural machine translation service from Amazon Web Services for fast, high-quality and affordable translation. It supports 75 languages and more than 5,550 language pairs, handles real-time and batch translation, and keeps document formatting intact. So it fits everything from live customer chat to high-volume document processing for global markets.

Amazon Translate does more than word-for-word translation. Instead, its deep learning models consider the full context of each source sentence to produce fluent, natural translations. Meanwhile, Custom Terminology keeps brand names, product names and industry terms consistent, and Active Custom Translation (ACT) lets you tune output with your own parallel data for domain-specific accuracy, without training or maintaining a custom model. Where an application must also speak that translated text aloud, Amazon Polly voices can turn it into natural-sounding audio.

Abilities Overview

75 languages
Supported Languages
5,550 pairs
Language Combinations
2M chars
Free Tier (Monthly)

Amazon Translate also works with the wider AWS ecosystem: S3 for document storage, Lambda for event-driven workflows, Comprehend for analyzing translated text and Transcribe for speech-to-text in multilingual pipelines. As a result, you can build complete multilingual workflows inside AWS. For scanned documents, Amazon Textract text extraction can pull the words off the page first so they can be translated.

Amazon Translate at a glance: 75 languages, 5,550 language pairs, 2 million free characters a month, and five best practices
Amazon Translate in three numbers, with five best practices.

Key Takeaway Amazon Translate provides neural machine translation across 75 languages, with custom terminology, formatting preservation and close AWS integration. If your organization needs to localize content, translate documents or run multilingual applications at scale, it is a quick route to production-grade translation on AWS.


How Amazon Translate Works

In practice, Amazon Translate runs as a serverless API. You send text in a source language, or let Translate detect it, name the target language and get the translated text back. Behind the API, in fact, neural machine translation models trained on a large, varied multilingual corpus read context and produce fluent output.

Real-Time Translation

For immediate needs, the synchronous TranslateText API handles up to 10,000 bytes of text per request and returns results in real time. As a result, this mode powers live chat translation, dynamic website localization and real-time content rendering. The real-time document translation API also translates whole documents (TXT, HTML, DOCX) and keeps the original formatting, so translated Word files retain their layout, fonts and structure.

Batch Translation

For high-volume work, the asynchronous TextTranslationJob API processes whole collections of documents stored in S3. First, you submit a batch job with the source and target languages and the input and output S3 locations, and Translate processes every file. The results then land in the output bucket. In addition, batch translation supports Word (DOCX), PowerPoint (PPTX), Excel (XLSX), plain text and HTML, which suits bulk localization of product catalogs, documentation libraries and marketing content.

Customization Features

Amazon Translate offers two ways to improve quality for domain-specific content:

  • Custom Terminology: Upload a glossary of terms that must translate a certain way, such as brand names, product names, acronyms and industry jargon. Translate then uses those terms exactly as you specified, overriding the model’s default. Custom Terminology costs nothing beyond standard per-character pricing.
  • Active Custom Translation (ACT): Provide parallel data, meaning examples of source and target translation pairs, with your batch translation job. Translate then uses it to customize the output at runtime, which improves accuracy for your domain without building, training, deploying or maintaining a custom model. ACT is especially useful for legal, medical, technical and financial content, where standard translations can miss domain nuance.

Core Amazon Translate Features

Beyond the translation APIs and customization options, several features make Amazon Translate suited to enterprise multilingual work:

Neural Machine Translation
Deep learning models consider the whole sentence to produce fluent, natural translations. As a result, they are much more accurate than older statistical or rule-based approaches.
Document Format Preservation
Translates DOCX, PPTX, XLSX, HTML and TXT files while keeping their formatting, layout, fonts and structure. So documents need no reformatting after translation.
Automatic Language Detection
Detects the source language automatically when it is not specified, which helps in pipelines where the input language varies. Behind the scenes, detection also relies on Amazon Comprehend.
Custom Terminology
Keeps brand terms, product names and industry jargon consistent. Also, glossary files cost nothing extra, and the terms apply during translation without changing how the model handles other content.
Active Custom Translation
Supply parallel translation examples to tune batch output for your domain. In turn, it improves accuracy for specialized legal, medical and technical content without custom models.
Profanity Masking
Optionally masks profane words and phrases in the output, replacing them with a generic string. For example, it is useful for content platforms and customer-facing applications that must filter inappropriate language.
Talk to an ExpertLocalizing content for new markets? Plan your translation pipeline with us.

Amazon Translate Pricing Model

Fundamentally, Fundamentally, Amazon Translate uses pay-per-character pricing with no minimum commitments. Rather than listing specific dollar amounts that change over time, here is how the cost structure works and what to watch for when estimating monthly spend:

Understanding the Costs

  • Standard translation: Charged per character processed, including whitespace and punctuation. The same per-character rate applies to real-time and batch translation, and pricing does not change with volume.
  • Active Custom Translation: Charged at a higher per-character rate than standard translation, to cover the runtime customization.
  • Custom Terminology: No extra charge, because the cost is included in the standard per-character rate.
  • Free tier: 2 million characters a month of standard translation for the first 12 months, which covers development, testing and modest production workloads.
Cost Optimization Strategy

To control costs, clean text before translation by removing unnecessary whitespace, normalizing formatting and stripping markup you do not need translated. Cache translated content so static text is not translated again. And when you know the source language, specify it, which skips the detection step. For current prices, see the official Translate pricing page.


Real-World Amazon Translate Use Cases

With 75 languages, Amazon Translate suits organizations with global content and communication needs. These are the use cases we implement most often for enterprise clients:

E-Commerce Product Localization
Translate product titles, descriptions, reviews and category pages for each target market. Batch translation handles whole catalogs, while Custom Terminology keeps brand names and product terms consistent.
Multilingual Customer Support
Translate customer chat messages in real time, so agents can support customers in any language. You can also add Amazon Transcribe for multilingual voice support and Comprehend for sentiment analysis of translated conversations.
Document Translation at Scale
Translate internal documentation, compliance materials, training manuals and legal documents while keeping their formatting. In addition, batch jobs can process thousands of documents at a time.
Website and Application Localization
Translate website content, application interfaces and user-generated content for international audiences. Real-time API integration translates content as it is created or requested.
Social Media and Content Analytics
Translate posts, reviews and comments from many languages into one language for unified sentiment analysis in Amazon Comprehend. As a result, global brand monitoring becomes possible across language barriers.
Healthcare and Legal Translation
Translate patient communications, medical documentation and legal filings, using Active Custom Translation for domain accuracy. Similarly, Custom Terminology keeps medical and legal terms correct and consistent.

Amazon Translate vs Google Cloud Translation

If you are comparing translation services across cloud providers, here is how Amazon Translate compares with Google Cloud Translation:

Ability Amazon Translate Google Cloud Translation
Language Support Yes: 75 languages, 5,550 pairs ✓ More languages
Neural Translation ✓ Context-aware NMT Yes: Neural Machine Translation
Custom Terminology ✓ Included at no extra cost Yes: Glossaries (Advanced API)
Custom Models Yes: Active Custom Translation (ACT) Yes: AutoML Translation
Document Translation ✓ DOCX, PPTX, XLSX, HTML, TXT Yes: PDF, DOCX, PPTX, XLSX
Batch Processing ✓ S3-based batch jobs Yes: Cloud Storage-based batch
Free Tier ✓ 2M chars/month (12 months) ◐ 500K chars/month
Ecosystem Integration Yes: S3, Lambda, Comprehend, Transcribe Yes: Cloud Storage, Functions, NL API

Choosing the Right Option

Both services offer strong neural machine translation, so your cloud ecosystem usually decides. On AWS, Translate’s native links to S3, Lambda, Comprehend and Transcribe make it the natural choice for multilingual pipelines. On Google Cloud, Cloud Translation integrates natively with Cloud Storage and the Natural Language API.

Google covers more languages, and its AutoML Translation offers more advanced custom model training. Amazon Translate, for its part, has a larger free tier (2M versus 500K characters a month), built-in profanity masking and ACT, which tunes translations at runtime without separate model training. Custom Terminology is also included at no extra cost, while Google’s glossary feature needs the Advanced edition.


Getting Started with Amazon Translate

Amazon Translate needs no setup: you call the API with text and get the translation back. The free tier gives 2 million characters a month for the first 12 months, which is enough for development, testing and modest production use.

Your First API Call

Here is a minimal Python example that translates English text into Spanish:

import boto3

# Initialize the Translate client
client = boto3.client('translate', region_name='us-east-1')

# Translate text
response = client.translate_text(
    Text='Welcome to our service. How can we help you today?',
    SourceLanguageCode='en',
    TargetLanguageCode='es'
)

print(f"Translated: {response['TranslatedText']}")
# Output: "Bienvenido a nuestro servicio. ¿Cómo podemos ayudarle hoy?"

For batch document translation, use the start_text_translation_job API to process whole S3 folders. To keep formatting in translated documents, use the translate_document API. For details, see the Amazon Translate documentation.


Amazon Translate Best Practices and Pitfalls

Advantages
75 languages with 5,550 translation pair combinations
Document translation preserves DOCX, PPTX, XLSX formatting
Custom Terminology included at no additional cost
Active Custom Translation enables domain-specific accuracy
Generous 2M character/month free tier for 12 months
Deep AWS integration with S3, Lambda, Comprehend, Transcribe
Limitations
Fewer languages than Google Cloud Translation
No volume discounts: one flat rate at any usage
Whitespace and punctuation count toward billable characters
ACT is charged at a higher rate than standard translation
Translation quality can be literal for some language pairs

Recommendations for Deployment

  • Define Custom Terminology before translating: Put brand names, product names and industry terms into a terminology file. After all, this one step prevents the most common error, a brand name being translated as an ordinary word.
  • Clean text to reduce character count: Remove unnecessary whitespace, strip HTML from non-HTML content and normalize formatting before submitting text. Every character, including spaces, is billed, so cleaner input costs less.
  • Cache translated static content: For content that rarely changes, such as product descriptions, UI strings and help pages, translate once and cache the result. Then you avoid paying to translate the same text on every request.
  • Use ACT for domain-specific quality: For legal, medical or technical content, provide parallel data through Active Custom Translation. In turn, it improves domain terminology and can reduce the post-editing that human translators need to do.
  • Combine Translate with Comprehend for multilingual analytics: Translate international content into one language, then analyze it with Comprehend for sentiment, entities and key phrases. Overall, that gives you unified brand monitoring and customer analytics across languages.

Key Takeaway Amazon Translate handles translation at scale across 75 languages, with neural accuracy, custom terminology and formatting preservation. To get the most from it, trim character counts, cache static translations and use Custom Terminology and ACT for domain quality. An experienced AWS partner can help you design multilingual pipelines that balance quality, cost and automation.

Our ServiceData, Analytics and AI

Frequently Asked Questions About Amazon Translate

Common Questions Answered
What is Amazon Translate used for?
Amazon Translate converts text from one language to another with neural machine translation. Common uses include e-commerce product localization, multilingual customer support chat, bulk document translation (Word, PowerPoint, Excel, HTML), website and app localization, social media analytics across languages, and healthcare and legal translation. It supports 75 languages and more than 5,550 language pairs.
Is Amazon Translate free?
Yes. The free tier gives 2 million characters a month of standard translation for the first 12 months. After that, standard translation uses flat per-character pricing with no volume discounts, Active Custom Translation costs more per character, and Custom Terminology is included at no extra cost.
What is the difference between Amazon Translate and Amazon Transcribe?
They do different jobs despite the similar names. Amazon Translate converts text from one language into another, for example English text into Spanish text. Amazon Transcribe converts speech into text, for example an audio recording into a written transcript. They often work together in multilingual pipelines: Transcribe turns spoken audio into text, and Translate then converts that text into the target language.

Technical and Quality Questions

How accurate is Amazon Translate?
Accuracy depends on the language pair and the subject matter. For common pairs such as English to Spanish, French or Portuguese, users generally report strong results. For specialized legal, medical or technical content, Custom Terminology and Active Custom Translation improve accuracy noticeably. Some users find translations too literal for certain pairs, so test with your own content and languages before going to production.
Can Amazon Translate preserve document formatting?
Yes. Real-time document translation keeps the original formatting of DOCX, HTML and TXT files. Batch processing also handles PPTX and XLSX files and keeps their layout, fonts and structure. So you avoid the manual reformatting that machine translation of formatted documents usually needs.

References

  1. 75 Languages, 5,550 Pairs, Real-Time and Batch Translation: AWS — Amazon Translate Developer Guide
  2. Per-Character Pricing and 2M Character Free Tier: AWS — Amazon Translate Pricing
  3. Google Cloud Translation Free Tier and Glossaries: Google Cloud — Cloud Translation Pricing
Newsletter

Security Insights, Delivered.

Practical, vendor-neutral analysis of threats, tools, and architecture decisions.

Book a Consultation

A practice principal will contact you within 1 business day.

Thank you! A Signisys principal will be in touch within 1 business day.