Piotr Dura

Piotr Dura has a diverse work history in the field of speech research and machine learning engineering. Piotr is currently employed as a Senior Speech Research Engineer at Sanas since October 2022. Prior to this role, Piotr worked at charactr from September 2020 to October 2022, where they held the position of ML Engineer | Researcher. Their responsibilities included training models in speech domain, conducting research on zero-shot TTS and any-to-any Voice Conversion models, as well as Speech Enhancement and custom Speech Vocoders.

Before joining charactr, Piotr worked at ByteDance as a Machine Learning Engineer from May 2019 to September 2020. In this role, they focused on low-level optimization of speech models such as Tacotron, Wavenet, LPCNet, WaveRNN, LAS, and RNN-T using TensorFlow, PyTorch, and Kaldi.

Prior to their time at ByteDance, Piotr worked at Samsung Electronics where they held the position of NLP | Machine Learning Engineer from July 2015 to April 2019. Their responsibilities at Samsung included the development of ASR for various devices, research and development of LM, G2P, and far-field noise-robust ASR. Piotr also worked on the development and optimization of novel WFST algorithms.

Piotr Dura pursued their Bachelor of Engineering in Computer Science from 2014 to 2018 at Uniwersytet im. Adama Mickiewicza w Poznaniu. Their field of study during this period was exclusively focused on Computer Science.