Principal TTS Researcher
Core
Designing and optimizing real-time Text-to-Speech (TTS) systems for automotive voice assistants, focusing on acoustic models, neural vocoders, and prosody prediction.
Role type
Principal TTS Researcher (Hands-on IC)
Builds
Production-grade TTS models and voice cloning solutions for car infotainment systems
Domain
Automotive voice assistants / Speech AI
Deliverable
production ML models
Required skills
TTS system development, C/C++, Python, PyTorch, TensorFlow, NLP, speech signal processing, linguistic tools, phonetic knowledge, transformer-based language models, autoregressive/non-autoregressive acoustic models, neural vocoders, model optimization (quantization, pruning, distillation), speech codecs, real-time streaming protocols, ONNX Runtime, TensorRT, TorchScript, zero-shot/one-shot/few-shot voice cloning, GPU/TPU cluster management
Preferred skills
Publications in INTERSPEECH/ICASSP/NeurIPS, tenure in world-level TTS/AI company or research institute, cross-functional leadership
Technologies
PyTorch, TensorFlow, Festival, Opus, MELP, ONNX Runtime, TensorRT, TorchScript
Responsibilities
Develop and optimize frontend and backend components of TTS systems; implement transformer-based models for prosody prediction; manage GPU/TPU clusters for model training and inference; deliver competitive TTS solutions for automotive clients
Seniority
Principal, hands-on IC