If you always do what you've always done, you'll always be where you've always been. - T.D. Jakes -
I used a smartphone and 3d-printed stethoscope to record heart sounds (an improved version of the previous stethoscope)
A simple demo video of how I measured my hearing threshold, which is the lowest level sound that can be heard.
Using bone-conducted (BC) sound and machine learning to authenticate the registered users and reject unknown ones.
I am developing a stethoscope attachment by which we can record heart sounds using our smartphone.
StethoPhone is a low-cost digital stethoscope made by using 3D printing technology and designed for use with smartphones.
I am developing an Android app that has some basic functions such as audio streaming, filtering, recording and file sharing.
This project aims at implementing a real-time sound localization using SVM, running on Raspberry Pi.
This project aims at implementing binaural speech enhancement method on a single board computer for a real-time use.
This project aimed at implementing "Spatial SpeechFocus" (check the previous video) on Android smartphone.
This project was a part of my team's homework in image processing course at Kumamoto University (1st-year master course).
This project aimed at developing a directional loudspeaker system that consist of only two loudspeakers mounted back to back.
We developed a localization algorithm that can distinguish the sound from the front & back; then, we experimented in an anechoic room with pink noise as a sound source.
ExperimentNormal Room - Pink Noise
We tested our sound localization algorithm in a normal room condition with pink noise as a sound source. Highest peak in the graph indicates the "true" direction.
ExperimentNormal Room - Speech Signals
We tested our sound localization algorithm in a normal room condition with speech signals as sound sources. Front-back confussions sometimes still occur.
We looked for a possibility to perform a cross-talk cancellation for bone conducted sounds. See the video; no sound, only animation.
I made a simple real-time sound localization using GCC-PHAT algorithm, not FDBM-based algorithm as written in the video.