Scholar
Rafael Valle
Google Scholar ID: SktxU8IAAAAJ
NVIDIA, UC Berkeley, CNMAT
Citations & Impact
Citations
2,640
H-index
16
i10-index
22
Publications
20
Co-authors
46
list available
Publications
Resume (English only)
Academic Achievements
- Co-invented Fugatto, Audio Flamingo, OMCAT, ETTA, Koel-TTS, P-Flow, the RAD* family of models, etc. Published papers include 'Fugatto: Foundational Generative Audio Transformer Opus 1'.
Research Experience
- Worked as a polymath research scientist and manager at NVIDIA, representing ADLR's (Applied Deep Learning Research) audio team. The team focuses on generative models with intelligence in audio understanding and synthesis, occasionally exploring vision. Was a Research Intern at Gracenote in Emeryville during Fall 2016, working on audio classification using Deep Learning. Previously, a Scientist Intern at Pandora in Oakland, investigating segments and scores that describe novelty seeking behavior in listeners.
Education
- PhD from UC Berkeley, advised by Prof. Sanjit Seshia and Prof. Edmund Campion; focused on machine listening and improvisation. Master's in Computer Music from HMDK Stuttgart, Germany; Bachelor's in Orchestral Conducting from UFRJ, Brazil.
Background
- Pursuing Superintelligence in Multimodal Generation and Understanding at Meta. Passionate about generative modeling, machine perception, and machine improvisation.
