Flitto Enhances Voice Data Collection for Big Tech

## Flitto Strengthens In-App 'Natural Call Data' Collection Infrastructure… Leads High-Quality Data Market for AI Learning Flitto, an AI language data specialist company, has significantly upgraded 'Arcade Talking Mission', its core voice data collection feature within its mobile app. This aims to stably supply high-quality data optimized for training artificial intelligence (AI) models such as speech recognition (STT) and natural language processing (NLP), by securing natural conversational data between actual users. This project was planned based on the demand from global big tech companies. The biggest feature is the direct use of call functions within the Flitto app without external app integration. This is the result of focusing on securing 'high-quality, natural' conversational data similar to real call environments. Users can communicate freely with only pre-set situations and keywords, capturing real-life conversation flows not restricted by scripts. Through this, vivid data close to daily life, rather than artificial conversations, is collected. This upgrade has unified the entire process of voice data collection and processing, from the complete internalization of in-app call functions to situation-based free conversation design, streamlined participation procedures, and the establishment of submission formats optimized for AI learning. In particular, Flitto's strength lies not just in simple collection, but in refining the collected data according to its own quality standards and directly performing 'meaning-based sophisticated labeling,' completing it in a form immediately usable for AI model development. The data is designed to reflect language-specific characteristics and usage environments, and initially, collection will begin with a focus on Korean and Japanese data. Subsequently, there are plans to gradually expand to various languages such as Arabic, Malay, Chinese, and English. The high-precision voice datasets completed in this way will undergo Flitto's rigorous quality inspection process and will be supplied to global big tech companies and major research institutions. Lee Jung-soo, CEO of Flitto, stated, "The upgrade of this Arcade Talking Mission has laid an important foundation for stably securing natural voice data reflecting real-world usage environments. Based on its expertise in independently carrying out the entire process from data refinement to labeling, beyond mere collection, Flitto will continue to build training data optimized for various languages and situations to contribute to the refinement and expansion of the global AI industry's utilization base." Flitto plans to continue its leading role in the AI training data market through its end-to-end data processing capabilities.