PhD in Speech & Language Tech: Audio Deepfake Detection

Aalto University; Espoo; On-site; EUR 32,000 - 38,000; Full time

Benefits offered by this job

Hybrid work model

Campus life Otaniemi

Job summary

Aalto University in Espoo, Finland invites applications for a Doctoral Researcher in Speech and Language Technology in the DICE group. The COWAMA project develops content-based watermarking for AI-generated audio and robust deepfake detection to address misinformation, copyright, and attribution.

You will build detectors for origin and authenticity, create diverse datasets, and contribute to open-source tools while publishing in top venues.

Qualifications

• Master’s degree or near completion in speech/audio processing, ML, signal processing, CS or E.E.
• Strong interest in doctoral research on audio deepfake detection, watermarking and generative AI.
• Good programming skills (Python and ML tools).
• Basic knowledge of deep learning, signal processing, speech/audio processing, or statistical ML.
• Interest in datasets, detector models, generative audio models, and evaluation.
• Motivation to publish scientific articles and contribute to open-source and reproducible research.
• Ability to work independently and in international teams.
• Fluency in English; Finnish not required.

Responsibilities

• Detect generated audio content from multiple sources and collaborate on robustness, evaluation, and generalization.
• Develop methods for detecting authenticity and origin of speech and audio content from various models and watermarking methods.
• Create datasets of real and generated speech and audio from different models and watermarking methods.
• Adapt and improve detector baselines for multi-objective detection (deepfake, content origin, watermark ID).
• Develop explainable, localized detection methods for watermark and deepfake attribution.
• Study robustness against signal-processing and adversarial attacks, and downstream processing like mixing.
• Publish research in leading journals and conferences; contribute to open-source software and reproducible outputs.

Skills

Doctoral research focus

Python

Machine learning tools

English proficiency

Experimentation

Independence and collaboration

Education

Master’s degree in related field

Tools

Python

Job description

Aalto University in Espoo, Finland invites applications for a Doctoral Researcher in Speech and Language Technology in the DICE group. The COWAMA project develops content-based watermarking for AI-generated audio and robust deepfake detection to address misinformation, copyright, and attribution.

You will build detectors for origin and authenticity, create diverse datasets, and contribute to open-source tools while publishing in top venues.