Markus Toman
Head Of Veritone Labs
Salzburg, Salzburg, Austria
Summary
👤
Senior🎓
Top SchoolMarkus Toman is a research-driven engineering leader with 11+ years building speech, audio and multimodal AI systems, currently heading Veritone Labs after leading voice engineering and R&D at VocaliD. He combines a PhD-level background in speech synthesis and medical informatics with hands-on experience shipping soft‑realtime, cross‑platform streaming TTS for mobile and cloud, and contributions to prominent open-source speech projects like Merlin. Markus has applied voice tech to improve accessibility for people with speech disorders and scaled voice talent monetization, while earlier work spans embedded systems, medical imaging, and large-scale web projects such as a national housing search engine. Known for bridging rigorous research and product delivery, he brings a practical focus on responsible, human-centered AI informed by clinical and academic experience.
11 years of coding experience
18 years of employment as a software developer
Vienna University of TechnologyDiplom-Ingenieur (MSc) Medical Informatics, Diplom-Ingenieur (MSc) Medical Informatics at Medical University of ViennaIng. EDVO/EDP, Ing. EDVO/EDP at HTL Wr. Neustadt
German, English
Stackoverflow
151reputation
45kreached
5answers
0questions
Github Skills (20)
python10
text-to-speech10
speech-synthesis10
keras10
deeplearning-ai10
deep-learning10
tensorflow10
voice-conversion9
theano9
machine-learning8
configuration-management7
parallel-processing7
script6
roblox6
internal-storage6
Programming languages (5)
Github contributions (5)
CSTR-Edinburgh/merlin
Nov 2016 - Feb 2018
This is now the official location of the Merlin project.
Role in this project:
ML Engineer
ML EngineerContributions:7 commits, 4 PRs, 41 comments in 1 year 3 months
Contributions summary:Markus contributed to the Merlin project, which focuses on deep learning for speech synthesis and voice conversion. Their work included fixing alignment paths related to HTK, updating speaker adaptation demos, and adding functionality to download datasets. They also implemented a layer freezing configuration parameter and added parallel acoustic composition capabilities to improve performance.
pythondeep-learningvoice-conversiontheanospeech-synthesis
m-toman/tacorn
Aug 2018 - Jul 2019
Combination of the Tacotron-2 implementation by Rayhane-mamah with the WaveRNN-inspired method by fatchord
Contributions:76 commits, 1 PR, 63 pushes in 10 months
combinationmethodtacotrontacotron-2wavernn
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.