Delightroom Enhances Sleep Analysis Accuracy Using RespiraSegNet to Overcome Noise

Sleep technology innovation company Delightroom has opened new horizons in the sleep tech market by unveiling a groundbreaking AI model, 'RespireSegNet.' This technology, announced at the International Conference on Electronics, Information, and Communication (ICEIC), is noteworthy because it has successfully resolved chronic problems faced by existing sound-based respiration analysis, namely errors caused by external noise or restless movements. RespireSegNet enhances OpenAI's powerful speech recognition AI 'Whisper' to enable stable and accurate respiration pattern analysis in any noisy environment. It has, in fact, demonstrated overwhelming accuracy in prediction experiments using sleep data from 287 individuals, and a lightweight version of this technology is already integrated into the 'Sleep Analysis feature' of Delightroom's flagship app, 'Alarmy,' reaching users. Alarmy's feature, which precisely detects various sounds such as breathing patterns and snoring during sleep using only a smartphone microphone and quantifies sleep quality, has become even more powerful with this upgrade. Shin Jae-myung, CEO of Delightroom, stated his ambition to continue leading innovation in the sleep tech field and consistently introduce useful features throughout the entire sleep process for users, now that precise sleep analysis has become possible with non-invasive AI technology. Building on this achievement, Delightroom plans to expand the application of its technology to include sleep apnea detection and solidify its competitive advantage in the digital healthcare and sleep tech sectors. To achieve this, the company is committed to its own technological development and paper publications, while also actively pursuing collaborations and investments with innovative partners such as AI mattress 'Sambunil' and digital healthcare 'Soundable Health.'