Senior Research Engineer - Voice

📋 Description

• Develop streaming and speech-to-speech systems for low-latency voice synthesis.
• Adapt models for conditioning inputs (emotion, speed, prosody, speaker control).
• Implement post-training optimizations (quantization, pruning, distillation) for real-time speech.
• Test novel architectures (neural codecs, diffusion, flow-matching) for realism.
• Define metrics for conversational speech, incl. latency-aware MOS.
• Apply DPO and distillation to fine-tune large-scale speech models.

🎯 Requirements

• Strong understanding of generative modeling for sequential or multimodal data.
• Experience with LLMs or transformer architectures.
• Proficiency in PyTorch, with distributed training and optimization.
• Time-series modeling and tokenization for audio or speech.
• Prototype quickly, test hypotheses, iterate efficiently.
• End-to-end DL model training from data prep to evaluation.
• Software engineering skills enabling contributions to large research infra.