Doctoral Researcher in Speech and Language Technology

About the role

Aalto University is looking for a Doctoral Researcher in Speech and Language Technology to join the funded project COWAMA: Content-based watermarking for AI-generated audio, led by Assistant Professor Lauri Juvela. The research takes place at the Department of Information and Communications Engineering (DICE) in Espoo, within a team working on deep generative models for speech and audio, speech synthesis, differentiable signal processing, deepfake detection and watermarking. The project develops methods for detecting the origin and authenticity of generated and watermarked audio, and evaluates robustness against removal, spoofing and adversarial attacks.

Key responsibilities

• Develop methods for detecting the authenticity and origin of speech and audio generated by multiple models and containing different watermarking methods.
• Create datasets of real and generated speech and audio from different generative models and watermarking methods.
• Adapt and improve detector baselines for multi-objective detection (deepfake detection, content origin detection, watermark identification).
• Develop explainable, localized detection methods for watermark and deepfake attribution, especially when generated content is mixed with other audio.
• Study robustness against signal-processing, generative resynthesis, adversarial and downstream processing attacks.
• Publish in leading journals and conferences and contribute to open-source software, models and reproducible research.
• Run deep learning experiments on high-performance computing infrastructure and evaluate with objective metrics and subjective listening tests.

Required profile

• Master's degree, or close to completing it, in speech or audio processing, machine learning, signal processing, computer science, electrical engineering or a related field.
• Strong interest in audio deepfake detection, watermarking, source tracing and generative AI.
• Interest in working with speech and audio datasets, detector models and experimental evaluation.
• Ability to work independently and collaboratively in an international research environment; fluency in English is required, Finnish is not.

Required skills

• Good programming skills, preferably Python and modern machine learning tools.
• Basic knowledge of deep learning, signal processing, speech/audio processing or statistical machine learning.

What we offer

• Starting salary of 3,143 EUR per month, increasing after a mid-term evaluation.
• Fixed-term position under the school's 2+2 model: two years initially, then two more after a successful review, with a six-month probation period; start in January 2027 or as agreed.
• Computing infrastructure including CPU/GPU clusters, access to CSC and LUMI, and the Aalto Acoustics Lab.
• Supervision by Assistant Professor Lauri Juvela and international collaboration with researchers in Japan, Spain and Sweden.
• Open-science approach, professional development and a hybrid work model with the main workplace at the Otaniemi Campus.