Senior Researcher, Speech Synthesis and Multimodal LLM
Core
Research and develop advanced speech synthesis and generation algorithms (e.g., TTS, voice conversion, sound/music generation) based on LLMs, multimodal/omnimodal LLMs for online applications.
Role type
Senior Researcher, Speech Synthesis and Multimodal LLM
Builds
Speech and audio synthesis systems for online applications
Domain
Gaming + Generative AI (Speech/Audio)
Deliverable
production ML models
Required skills
Speech/audio processing, modern generative models (diffusion, flow matching, autoregressive, codec-based), LLM extension to speech/audio modalities, full-duplex/streaming spoken dialogue systems, Python, deep learning frameworks (PyTorch), distributed training
Preferred skills
Experience with speech tokenizers, multimodal adapters, speech-text joint training, real-time interaction modeling
Technologies
PyTorch, LLMs, diffusion models, flow matching, autoregressive models, codec-based approaches
Responsibilities
Research and develop speech synthesis algorithms, optimize synthesis systems for effectiveness and scalability, explore full-duplex/streaming multimodal LLM capabilities, collaborate cross-functionally from prototyping to production
Seniority
Senior, hands-on IC