Senior Generative Speech AI Scientist (Hybrid)

Spotify Los Angeles; Greater London; Hybrid; GBP 110,000 - 150,000; Full time

Benefits offered by this job

Flexible work locations

Six months parental leave

Employee assistance program (The Feels

GreenHouse learning opportunities

Job summary

Spotify is seeking a senior applied research scientist to develop novel ML techniques and architectures for state-of-the-art speech-to-speech models, pushing the boundaries of production-ready speech technology.

The role is based in London or Stockholm with hybrid flexibility, collaborating across engineering and data teams to scale research into production and explore new ideas for quality, realism, and broader market use cases.

Qualifications

• PhD in ML or related field required.
• Experience with transformers, GANs, diffusion models, flow matching, VAEs, or audio codecs.
• Experience in developing generative models for speech synthesis, speech recognition, audio/music, NLP, or computer vision.

Responsibilities

• Develop and experiment with new methods for speech synthesis and speech recognition, building on the latest research.
• Expand speech use-cases targeting different markets and products.
• Collaborate with engineering and data teams to improve infrastructure and data quality for speech pipelines.
• Share knowledge and best practices with other researchers within Speak.
• Turn proven ideas into scalable production-ready components.

Skills

Python

Transformers

GANs

Diffusion models

Flow matching

VAEs

Audio codecs

Education

PhD in ML or related field

Tools

PyTorch

Job description

Spotify is seeking a senior applied research scientist to develop novel ML techniques and architectures for state-of-the-art speech-to-speech models, pushing the boundaries of production-ready speech technology.

The role is based in London or Stockholm with hybrid flexibility, collaborating across engineering and data teams to scale research into production and explore new ideas for quality, realism, and broader market use cases.