Research Scientist - Generative Audio

Stack

Responsibilities

• ▹Conduct groundbreaking research in generative audio using diffusion or flow matching models
• ▹Research vocal and speech synthesis, ML-based audio processing and signal processing
• ▹Research post-training techniques for music generation: preference alignment (DPO, RLHF, KTO), reward model design, reinforcement learning
• ▹Research iterative music generation and audio editing: stem replacement, instrumentation change, mood changes, tempo and structure changes
• ▹Run large-scale experiments on Spotify's infrastructure with 700 million monthly active users
• ▹Create practical applications that harness generative technologies
• ▹Collaborate with scientists, engineers, product managers, designers and analysts
• ▹Have direct impact on Spotify's products, tools and services
• ▹Publish findings, deliver talks, attend conferences

Requirements

• ▹Ph.D. in Computer Science, Mathematics, Engineering or related field
• ▹Experience in one or more of: generative modeling, machine learning, music information retrieval, speech processing, audio processing, signal processing, probabilistic modeling, computer vision

Nice to have

• ▹Previous industry experience

Soft skills

About the company

Spotify's Artist-First AI Music Lab pioneers generative music technologies that put artists and songwriters first, with fair compensation and transparent credit.