腾讯游戏-语音大模型-高级研究员
Core
Research and develop next-generation voice synthesis base models to enable intelligent NPCs in games to interact with players in real-time like real humans.
Role type
Senior IC machine-learning researcher (voice synthesis)
Builds
Real-time, highly human-like voice synthesis for game NPCs
Domain
Gaming + Audio AI
Deliverable
production ML models
Required skills
Large model post-training, model distillation, neural audio codecs, voice tokenization, zero-shot voice cloning, emotional and prosody control, streaming low-latency synthesis, distributed training, reinforcement learning (RL/RLVR), TTS evaluation pipelines, online inference optimization
Preferred skills
Game engine integration, cross-functional collaboration, patent/paper publication
Technologies
Python, PyTorch, DeepSpeed, Megatron-Core, FSDP, vLLM, SGLang, EnCodec, DAC, RVQ, HiFi-GAN, Vocos, CosyVoice, MiniMax Speech, Seed-TTS, Unreal Engine
Responsibilities
Research frontier TTS technologies for intelligent NPC dialogue; Develop new voice synthesis base models; Build and curate NPC voice generation datasets; Collaborate with game engine teams for integration; Track and apply latest advancements in voice synthesis and game AI
Seniority
Senior, hands-on IC
