Anatomy of Industrial Scale Multilingual ASR

arXiv:2404.09841v1 Announce Type: cross
Abstract: This paper describes AssemblyAI's industrial-scale automatic speech recognition (ASR) system, designed to meet the requirements of large-scale, multilingual ASR serving various application needs. Our system leverages a diverse training dataset comprising unsupervised (12.5M hours), supervised (188k hours), and pseudo-labeled (1.6M hours) data across four languages. We provide a detailed description of our model architecture, consisting of a full-context 600M-parameter Conformer encoder pre-trained with BEST-RQ and an RNN-T decoder fine-tuned jointly with the encoder. Our extensive …

abstract application arxiv asr assemblyai automatic speech recognition cs.cl cs.lg cs.sd data dataset diverse eess.as industrial languages multilingual paper recognition requirements scale speech speech recognition training type unsupervised