PhD Research Intern: Reference-Free Audio Quality Assessment for Generative Audio (m/f/d)

Location: Nürnberg, Germany

Preferred start: Late January 2027 (or later).

Team: Sound Experiences Lab, ATG

About the Role

Generative audio systems increasingly produce realistic content, but their outputs may contain technical, non-semantic degradations such as tonal instability, transient smearing, temporal discontinuities, and spatial inconsistencies.

As a PhD Research Intern, you will investigate reference-free machine-learning methods for predicting the technical quality of generative audio. You will build on Dolby’s research in perceptual audio quality assessment and explore how knowledge learned from conventional and neural audio codecs can be transferred to generative content.

What You Will Accomplish

• Analyze technical artifacts in generative audio.
• Assess the suitability of available subjective quality data.
• Establish baselines using existing reference-free quality metrics.
• Adapt an existing perceptual audio-quality model using transfer learning.
• Evaluate generalization to previously unseen generative systems.
• Develop a research prototype and present the findings.

Qualifications

• Currently pursuing a PhD in machine learning, computer science, electrical engineering, audio signal processing, or a related field.
• Experience with deep learning and machine-learning experimentation.
• Strong programming skills in Python.
• Experience with PyTorch or a similar deep-learning framework.
• Strong analytical, experimental, and communication skills.
• Ability to work independently and collaborate effectively in a research environment.
• Experience in perceptual audio quality assessment or generative audio would be an advantage.