Adoram Erell

Experience

• 

  Tel Aviv University, EE department

  Sept 1985 - Sept 1987

  Post Doctoral fellow

  I gave up on nuclear physics and looked for ways to transition into engineering. I was given an opportunity in the EE department to do research on modelling of the auditory system, on methods to exploit this model in designing speech quantization schemes, and speech enhancement. During this period I taught myself signal processing, and taught undergraduate courses of electromagnetic fields and signal processing.

• 

  SRI International

  Sept 1987 - Sept 1989

  International Fellow

  After presenting a paper on auditory-based speech compression at an international conference, I was offered a fellowship with SRI’s speech recognition research team for one year. I succeeded to improve the noise-robustness of their system, and they extended my fellowship to a second year.

• 

  Israeli Aircraft Industry

  Sept 1991 - Apr 1994

  Rsearcher (Alef Mekhkar)

  Developing signal processing algorithms for a homing RADAR

• 

  DSPC

  Apr 1994 - Dec 1999

  Head of speech group

  I headed a small team which developed a speaker-dependent Voice Recognition system. It was first implemented in a stand-alone chip, then in the mobile chips for a name-dialing application. I was doing the algorithm and supervising the implementation. I also developed a single-mic noise suppressor and echo canceller for the mobile chips.

• 

  Intel Corporation

  Jan 2000 - Jan 2005

  Manager, speech recognition R&D

  When Intel acquired DSPC I was encouraged to upgrade the VR capability to a speaker-independent system for text-based name-dialing. Five experienced researchers from Converse joined the team and we ended up with a very efficient name-dialing application running on Windows Smartphone. It supported 6 languages and the performance with a vocabulary of 100 names was promising. However, at that point several third-parties emerged with similar solutions, wider language coverage, and more reasonable business models. Intel decided to close the project. Show less

• 

  Marvell Israel Ltd.

  Apr 2005 - Dec 2015

  In Dec 2010 I have been called up to "rescue” the audio solution (which had been frozen for 10 years) during a crisis with a customer. After solving the crisis, we realized at Marvell that the audio solution needs a major upgrade and I gradually phased out from the modem algorithms and became the leading developer and owner of the voice enhancement algorithms in Marvell’s mobile division. I have crafted the algorithms down to fixed-point C compatible with the DSP compiler, and this code was integrated on target by the FW team. I also supervised the audio lab team which validated the algorithms on customers’ devices and internal prototypes, and supported their tuning to different products. Show less After the VR project had been closed, I joined the modem algorithms research team. Initially I worked on a WCDMA advanced receiver, and then I joined the LTE sub-team to work on the emerging new LTE standard. I had two responsibilities: (1) I developed and owned the initial-search algorithm, the neighbor-cell search and measurements, and AFC. I developed them from scratch down to fixed-point for FW and HW implementations. These algorithms are running today in the LTE REL8 and REL9 Marvell products. (2) I represented Marvell in the 3GPP LTE RAN1 (physical layer) meetings. Initially I used these meetings to familiarize myself and the team with the subtleties of the evolving standard and influence the algorithm and system development. Later I took a more active role in the meetings, contributing to the MIMO-feedback specifications. During that phase I was leading two PhD’s from Marvell Santa Clara, who maintained a system-level-simulation. Out of that activity came out around 20 patents, several of which are essential to the LTE standard. Show less

  • Speech Enhancement Algorithms

    Jan 2011 - Dec 2015

  • LTE physical layer algorithms

    Apr 2005 - Dec 2010
• 

  DSP Group

  Jan 2016 - now

  Senior Algorithm Engineer

  Focal point and main innovation engine in DSPG's "Smart Voice" algorithm team.

• 

  Afeka Center for Language Processing

  Jan 2016 - Nov 2018

  Senior Researcher
• 

  Commusicator Inc

  Oct 2018 - now

  Co founder and CTO

  Performing as CTO in a small (team of 3) startup company, personally doing all IP development for Solfy, a web application for Solfege learning and practicing.