- A Neural Machine Translation service that provides fluent translation with higher accuracy than traditional statistical and rule-based translation models.
- Powered by deep learning models that deliver fast, accurate, and affordable translation between supported languages.
- Supports batch translation, real-time, and on-demand translations.
Common Use Cases
- Language localization
- Translating content for text analytics
- Cross-lingual communication
- Amazon Translate uses a Translation Model that consists of two (2) components:
- Read a source text one word at a time.
- Creates a semantic representation of each word.
- Uses the constructed semantic representation of the encoded text and translates it to the target language one word at a time.
- The decoder also uses a method called the Attention mechanism to translate obscure words or phrases correctly.
- When Automatic Language Detection is enabled, Amazon Translate uses Amazon Comprehend on the backend to automatically detect the language used in the source text.
- Broad Language Coverage
- Neural-Network Based
- Produces high-quality translations by utilizing deep learning technologies.
- Customized Machine Translation
- Enables you to tailor your translations to suit domain-specific terminologies.
- Useful for changing the tone, style, and word choices for the translated text.
- Secure Machine Translation
- Communications between Amazon Translate and your application is SSL-protected.
- Standard Translation
- Billed for $15.00 per million characters
- Active Custom Translation
- Billed for $60.00 per million characters