Skip to main content Skip to navigation

SALAD: Smart AI Language Assistant Daily

Project Overview

The document discusses SALAD, an innovative AI-driven application aimed at enhancing Japanese language learning for non-native speakers. This app incorporates a range of features such as translations, speech recognition, vocabulary tracking, grammar explanations, and the unique ability to generate personalized songs based on users' vocabulary, creating an engaging and customized learning experience. SALAD addresses the common challenges learners face when attempting to communicate with native speakers, thereby fostering greater confidence and skill development. User feedback suggests a strong belief in the app’s potential to significantly improve language proficiency, although it also highlights some challenges, including issues related to language specificity and dependence on external APIs for functionality. Overall, the document underscores the transformative role of generative AI in education, particularly in language acquisition, by providing tailored, interactive, and effective learning tools that cater to individual user needs.

Key Applications

SALAD (Smart AI Language Assistant Daily)

Context: Language learning for foreigners in Japan, targeting beginners and intermediate learners.

Implementation: The app integrates modules for translation, vocabulary tracking, grammar explanations, and song generation. Users can input text or speech, receive translations, and track progress.

Outcomes: Users reported enhanced confidence in their language skills, with 60% expressing belief in SALAD's effectiveness. The app offers personalized learning experiences and integrates entertainment through music.

Challenges: Limitations include language specificity (only English to Japanese), reliance on third-party APIs like ChatGPT, and partial success in song generation.

Implementation Barriers

Technical Limitations

The system is limited to English to Japanese translation, restricting its user base.

Proposed Solutions: Potential to expand language pairs to include more languages.

Dependency Issues

Heavy reliance on the ChatGPT API affects performance and availability.

Proposed Solutions: Developing a custom fine-tuned language model could enhance control and capabilities.

Quality of Output

Partial success in song generation may not meet user expectations.

Proposed Solutions: Further research and development are needed to improve song quality and variety.

Project Team

Ragib Amin Nihal

Researcher

Tran Dong Huu Quoc

Researcher

Lin Zirui

Researcher

Xu Yimimg

Researcher

Liu Haoran

Researcher

An Zhaoyi

Researcher

Kyou Ma

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Ragib Amin Nihal, Tran Dong Huu Quoc, Lin Zirui, Xu Yimimg, Liu Haoran, An Zhaoyi, Kyou Ma

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies