Within Personalization, the Speak Team owns the development of Spotify's state-of-the-art speech models, contributing to speech recognition, speech synthesis, and speech-to-speech models. We craft voice models that match human-level emotional expressiveness, so we can deeply engage our listeners and support creators at scale. Our groundbreaking work on speech synthesis relies on state-of-the-art deep learning methods and evaluation techniques, highly efficient data processing and model serving, and capturing audio of outstanding quality from our voice talent pool.
We're looking for a senior applied research scientist with experience in developing novel ML techniques and architectures and with a strong interest in working across a full production pipeline to produce state-of-the-art generative conversational speech-to-speech models. You'll collaborate with our engineering teams to help develop our production pipelines, explore new ideas and methods to improve quality, understanding and realism, as well as push the frontiers of what is possible with our speech technology.
What You'll Do- Develop and experiment with new methods for speech synthesis and speech recognition, along with end-to-end approaches, building on the latest research and ideas.
- Work towards the expansion of our speech use-cases targeting different markets and products.
- Be part of a highly motivated research team dedicated to building and creating models at scale to power the Spotify platform.
- Champion best practices for research and development, sharing your knowledge and experience with other researchers within Speak.
- Collaborate with our engineering and data teams on ideas requiring new infrastructure or new high-quality data, as well as to help improve our speech recognition and speech synthesis pipelines, and help turn proven ideas into scalable products.
Who You Are- You have a strong background in ML (PhD degree on top of professional experience), and
- experience in working with any of the following: transformers, GANs, diffusion models, flow matching, VAEs, audio codecs.
- You have experience in developing generative models for speech synthesis, speech recognition, audio/music, natural language processing, or computer vision.
- You have strong experience with Python, particularly PyTorch.
- You have strong communication skills and the ability to explain technical ideas with clarity to technical and non-technical people alike.
- You have experience in an academic or professional setting conducting high-quality research.
Where You'll Be- This role is based in New York City.
- We offer you the flexibility to work where you work best! There will be some in person meetings, but still allows for flexibility to work from home
The United States base range for this position is $169,157 - $241,653 plus equity. The benefits available for this position include health insurance, six month paid parental leave, 401(k) retirement plan, a monthly meal allowance, 23 paid days off, 13 paid flexible holidays. These ranges may be modified in the future.
