微信-语音识别算法工程师
Core
Develop and optimize low-latency, high-accuracy streaming speech recognition models for real-time dialogue, voice assistants, and hold-to-talk scenarios.
Role type
Senior IC speech recognition algorithm engineer (streaming ASR)
Builds
Streaming ASR models, speech understanding models, and complex scenario recognition capabilities
Domain
Consumer technology / Speech AI
Deliverable
production ML models
Required skills
Streaming ASR model architecture, CTC/Transducer/Conformer/Transformer, speech understanding, hotword enhancement, data cleaning, PyTorch, model training, latency optimization
Preferred skills
Multilingual/multidialect recognition, end-point detection, punctuation prediction, large-scale data curation
Technologies
PyTorch, Streaming Conformer, CTC, Transducer, Transformer
Responsibilities
Optimize model architecture and decoding strategies for streaming scenarios; enhance speech understanding for context and intent; improve recognition in complex scenarios (noise, overlapping speech, accents); optimize post-processing pipelines (punctuation, correction); manage large-scale data curation and labeling; establish evaluation metrics and drive model deployment.
Seniority
Senior, hands-on IC