Deep Neural Network

Phone calls in crowded places often fail for one simple reason: the person on the other end hears the noise before the voice. Samsung is taking aim at that problem with Galaxy Buds4 Pro, a pair of earbuds designed to keep speech clear even when the surroundings are far from quiet. The system behind that promise is not a standard noise reduction setup. Samsung uses next-generation Sensor Fusion technology, which is meant to keep the user’s voice prominent against traffic, chatter, and wind. Three microphones and bone conduction work together Wireless earbuds usually struggle because the microphones sit far from the mouth. That distance makes it easier for background noise to overpower speech, especially when a device relies on a single sensor. Galaxy Buds4 Pro addresses that limitation with three high-performance microphones. Two external microphones capture sound direction directly, while one internal microphone helps read the audio that travels through the ear canal. Samsung also adds a bone-conduction-based Voice Pickup Unit. This component reads physical vibrations from the head while the user is speaking, then combines that data with the other audio signals to rebuild the voice more completely. Audio Input System Role Benefit 2 external microphones Capture direction of sound Helps isolate the user’s voice from outside noise 1 internal microphone Reads sound through the ear canal Improves voice reconstruction Bone-conduction Voice Pickup Unit Detects head vibrations Adds another layer of voice detection AI processing that stays on the device At the center of the audio processing is a Deep Neural Network algorithm. The artificial intelligence system is built to separate the user’s voice from surrounding noise in real time. Samsung says the challenge was fitting that capability into a small earbud without wasting power. The company addresses that by using on-device processing that reduces workload by 10 percent and cuts the data model size to 30 percent of its original size. That approach keeps processing local to the earbuds instead of pushing everything elsewhere. As a result, the system can analyze audio in real time and capture sound detail up to 16 times richer than the previous generation. Designed to sound natural during calls The goal is not just to reduce noise, but also to prevent the user’s voice from sounding flat or robotic. Samsung says Galaxy Buds4 Pro is tuned to preserve the original character of speech, including consonant emphasis and higher tones. That matters most when the background is busy and unpredictable. With real-time analysis, the person on the other side of the call is supposed to hear speech that remains natural and easy to understand. Adaptive when the fit changes Another common problem appears when users walk, chew, or move more actively. Small shifts in the earbuds can create fit leakage, allowing outside sound to enter again through tiny air gaps. Galaxy Buds4 Pro monitors the balance between internal and external microphones automatically. When a leak is detected, the system adjusts audio processing dynamically within milliseconds so call quality stays stable. For users inside Samsung’s ecosystem, there is an added boost. When connected to a Galaxy phone, the earbuds can activate Super Wideband up to 16 kHz for a fuller and clearer sound. Tested in genuinely harsh environments Samsung did not rely only on light simulations during development. The company brought extreme real-world conditions into the lab with a giant wind generator to test call performance. Field testing also took place in difficult locations, including noisy shopping centers, high-roof train stations, windy outdoor areas at night, and car cabins with the windows fully open. Those tests show how physical limits in earbuds can be offset by Sensor Fusion, on-device AI processing, and richer voice capture. That combination is what helps Galaxy Buds4 Pro stay clear when calls happen in the middle of real-world chaos. Source: www.gadgetdiva.id

Galaxy Buds4 Pro Keeps Calls Clear, Even in Crowds and Strong Wind