Conduct generative audio research using diffusion or flow matching models.
Focus on vocal synthesis, including speech synthesis and related ML-based audio and signal processing.
Focus on post-training, including preference alignment (DPO, RLHF or KTO), reward model design and reinforcement learning for output quality and controllability.
Focus on editing, including iterative music generation and audio changes such as stem replacement, instrumentation, mood, tempo and structure.
Run large-scale experiments on Spotify's infrastructure.
Build practical applications of generative technology for listening experiences.
Collaborate with scientists, engineers, product managers, designers, user researchers and analysts.
Influence Spotify's products, tools and services.
Publish findings, give talks and attend top research conferences.
Requirements
Ph.D. in computer science, mathematics, engineering or a related field; industry experience is helpful.
Experience in generative modeling, machine learning, music information retrieval, speech, audio or signal processing, or a related area.
Deep expertise in at least one focus area: vocal/speech synthesis, post-training alignment (PPO, GRPO or DPO), or audio-to-audio generation and text-guided editing.
Publications at leading venues such as ICASSP, ISMIR, INTERSPEECH, ICLR, AAAI, IJCAI, NeurIPS, ICML, CVPR, ECCV or ICCV.
Strong coding skills in Python, PyTorch and NumPy.
Creative problem solver who wants to build products that add value for millions of people.
Motivated to turn research ideas into products that operate at scale.
Ability to explain complex topics simply and build strong working relationships.
Must be located in an EMEA country where Spotify has a work location; team works on CET/GMT time.
Done for you
We'll apply to this for you
Denidel's HR team submits and tracks the application on your behalf.
