Bengaluru's Sarvam AI Launches Mayura to Improve English-to-Indic Language Translation
Bengaluru-based startup Sarvam AI has introduced a new translation model known as Mayura to tackle the enduring difficulties in English-to-Indic language translation, particularly in everyday, informal communication.
Enhancing Real-world Language Patterns
The Mayura model is designed to focus on real-world language patterns such as code-mixing and colloquial expressions. Its primary goal is to provide translations that are not only accurate but also more relatable for Indian multilingual speakers. The model facilitates translations among 10 major Indian languages, including Hindi, Tamil, Telugu, Malayalam, Punjabi, Odia, Gujarati, Marathi, Kannada, and Bengali.
Revolutionizing Translation Models
Traditional translation models have faced challenges in capturing the nuances of how Indians communicate due to the blending of regional languages with English. Sarvam Translate, on the other hand, adopts a unique approach. It leverages diverse real-world data for training the model, ensuring that regional dialects, slang, and code-mixed phrases are preserved in translations. This breakthrough is poised to enhance the accessibility of digital content, social media, and e-commerce services for millions of users across India.
Addressing Gender-specific Language
Sarvam Translate also addresses gender-specific language in Indic languages, a crucial area where traditional models often fall short. By incorporating a "gender toggle" for first-person translations, the model ensures appropriate gender representation in conversations, whether with AI-powered voice chatbots or human agents.
Domain-specific Translations
Besides facilitating casual conversations, Mayura is tailored for domain-specific translations in areas like legal, scientific, and technical documents. This ensures that complex terminology is accurately rendered in simple, accessible Indic language. The dual-stream architecture of Sarvam AI's model preserves formatting elements in technical content, making it particularly beneficial for educational and government communications.
Availability and Integration
Mayura is now accessible as an API, enabling developers and businesses to seamlessly integrate it into their applications. As part of Sarvam AI's suite of products, Mayura complements other tools such as Sarvam Agents, Sarvam 2B (an open-source language model), and Shuka 1.0 (an audio language model).
