TL;DR — I am Robin Scheibler, a research engineer at Google Deepmind working more often than not on (neural) audio processing and stuff. I like coding and making hardware, and I like it open source.
Presentation at Interspeech 2024 about UNIVERSE++, a diffusion-based universal speech enhancement model with high content preservation (work done at LY Corp.).
I have joined Google Deepmind.
I gave a high level presentation of our work on optimization for blind source separation during LINE DEV DAY 2020. Watch the video.
Autumn will be busy. We will present our paper Indpendent Vector Analysis with more Microphones than Sources at WASPAA in October. The code is already available. In addition, Daiki Horiike will present “Blink-former: Light-aided beamforming for multiple targets enhancement” at MMSP, and Kouei Yamaoka will present “Sub-Sample Time Delay Estimation via Auxiliary-Function-Based Iterative Updates” at WASPAA too.
We presented our work on using blinkies for source separation and acoustic event detection at the SIP symposium in Nagasaki.
High performance algorithms leveraging structural properties of audio signals in reverberant environments.
I apply techniques from information theory and theoretical computer science to derive low-complexity algorithms for signal processing.
Reproducible computations, DIY devices, and citizen measurements of radiation and Arsenic in water.
Some random ramblings mostly related to building physical prototypes and tinkering with various things.
December 12, 2017
I recently started playing with the ESP8266, which was nice but somewhat limited, and now its older brother, the ESP32. Given their price and their I2S module, they make very nice platforms to start playing with audio processing and distributed microphone arrays. Just get an SPH0645 I2S microphone and get started! Amazing.
February 04, 2014
I have recently been interested in the ATtiny85 microcontroller. Its minimalistic, yet powerful features make it a very attractive platform for simple hacks and gadgets. In addition, I had been interested in learning the technique of charlieplexing, i.e. controlling many LEDs with very few pins. The ATtiny85, with its 5 GPIO pins available by default (sacrificing ISP yields an extra pin, but I didn’t want to go down that road), seemed like the perfect candidate.
February 13, 2013
Practical information on basic algorithms might be sometimes challenging to find. In this article, I break down two fundamental algorithms to compute the discrete Fourier transform (DFT, inverse transform is iDFT) of real-valued data using fast Fourier transform algorithm (FFT/iFFT).
©2018 Robin Scheibler. CC-BY-SA 4.0
