Conversational AI is an AI technology that naturally communicates with humans and is rapidly developing in various interactive services such as chatbots, voice assistants, and customer support systems. In particular, the importance of on-device conversational AI is growing as the prevalence of smart devices increases and the demand for personalized AI services increases . The Brain-Audio team is intensively researching and developing speech AI for on-device conversational AI solutions.
Brain-Audio recently collaborated with Qualcomm, a global semiconductor company, and successfully implemented Conversational AI technology (STT+LLM+TTS) in an affordable AI chip costing 100,000 won.
The technology was unveiled at CES 2025 atboth our booth and Qualcomm's booth, making MindAI an official Qualcomm partner. This collaboration marks an important milestone that brings us one step closer to democratizing high-performance on-device AI technology.
마음AI, 올해 CES서 퀄컴과 협업성과 최초 공개 성료 - 머니투데이
마음AI, 퀄컴 'IoT 엑셀러레이터 프로그램' 공식 멤버 등록
The Brain-Audio team is working on commercializing ASR-TTS-LLM as a single end-to-end system, going beyond the step-by-step module approach(ASR → language model → response generation → TTS) used in existing voice conversation systems.
The existing system suffers from information loss between modules, increased latency, and difficulty in optimization. To solve these problems, we are looking to build a single model that can be trained end-to-end, and can immediately understand and respond to voice input.
With this integrated approach, we plan to deliver a richer, smoother user experience.
We are looking for talented individuals to grow and challenge themselves in various research areas such as on-device AI, end-to-end voice conversation models, and AI Call Center (AICC) solutions.
<aside> 📝 How to apply