Alessio Falai

Portrait of Alessio Falai

Hi, I’m Alessio, a PhD student at the Music Technology Group of Universitat Pompeu Fabra in Barcelona, where I work on music and audio representation learning.

Before that I was an Applied Scientist at Amazon, taking speech generation models for Alexa and beyond from research to production, with a keen interest in multilingual and multimodal systems.

My story

My story

From my first lines of C to a PhD in Barcelona. Scroll to move along the line.

  1. Falling in love with C

    Self-taught programmer

    In middle school I fell in love with the words of Kernighan and Ritchie in their book on the C programming language. From then on I kept experimenting with programming and developed a lasting passion for Linux and open-source software.

  2. ICT diploma at I.T.S. Tullio Buzzi

    High-school diploma in ICT, 100/100

    Five years of network systems, computer architectures and programming basics, finished with a 100/100 diploma and a final project implementing a C++ holographic experience. In my 4th year I interned at Develer in Florence (June 2015), where I built a library and GUI to read, generate and print barcodes.

  3. Bachelor's degree in Computer Science at UNIFI

    BSc in Computer Science, 110/110 with honours

    My bachelor’s focused on the mathematical and logical foundations of computer science, advanced programming and software design, and computer architectures. I graduated with 110/110 with honours and a thesis on refining a Python library to perform OCR on identity documents.

  4. Freelance backend and website work

    Backend Developer and Website Administrator · Freelance

    Alongside my studies I took freelance work, starting with backend features for EasyClaims, an insurance-claims video verification interface. I then administered four WordPress and WooCommerce sites, from fashion e-commerce shops to a textile catalog and a photography portfolio.

    tognonishop.comstudio54italy.itloscoiattolofirenze.itwalterquiet.com

  5. Computer vision and OCR in industry

    Software Developer · QI-Lab, Continental, Quid Informatica

    At QI-Lab I built an OCR library for identity documents, with custom edge detection and image binarization as pre-processing, while also delivering a modular Qt GUI that operators at Continental use to log visual inspections of factory parts. At Quid Informatica I then shipped three production Python micro-services for banks that classify identity documents and extract their text, using NER to parse raw OCR output and a text classifier to categorize financial documents.

    Image binarization blog post (QI-Lab)Quid Informatica product video

  6. Master's degree in Artificial Intelligence at UNIBO

    MSc in Artificial Intelligence, 110/110 with honours

    My master’s covered the statistical, logical and mathematical foundations of AI, machine and deep learning, knowledge representation, computer vision and NLP. Along the way my team tackled the NeurIPS 2020 Flatland challenge, and I graduated with 110/110 with honours and a thesis on conditioning text-to-speech synthesis on dialect accent.

    Thesis audio samplesFlatland challenge write-up

  7. Speech generation at Amazon

    Applied Scientist

    I joined Amazon in 2021 as an intern, working remotely from Italy, and came back full-time that November to take multilingual text-to-speech models from research to production.

    • 2021As an intern, built forecasting models for warehouse capacity planning during COVID-19.
    • 2021–22Released Alexa male voices in more than 10 locales. Corriere della Sera (it-IT)
    • 2022–23Developed more than 60 cross-lingual on-device voice models, deployed in BMW and Mini cars. AnnouncementVideo
    • 2023–24Led a team of 3 scientists delivering LLM-based speech generation for ads.
    • 2024–25Trained multi-modal LLMs for the releases of Alexa+ and Amazon Nova Sonic. Alexa+Nova Sonic
    • 2025–26Led a unified inference implementation for our conversational system, and the launch of Alexa+ in 7 countries, from the UK (March 2026) to Brazil (June 2026). UK launchBrazil launch

    Publications

  8. PhD at the Music Technology Group

    PhD student · Universitat Pompeu Fabra · “la Caixa” Foundation doctoral fellow

    In November 2026 I left Amazon and moved from London to Barcelona to start a PhD in music and audio representation learning at the MTG of Universitat Pompeu Fabra, supported by a “la Caixa” Foundation doctoral fellowship. My supervisor is Xavier Serra and my co-supervisor is Frederic Font.

    MTG announcementAcademia