Mousa

WORK HISTORY

EDUCATION

ABOUT MOUSA

I am a senior research engineer working in the field of machine learning, namely deep learning technology powering Apple Siri, the Apple\'s ASR based smart assistant. In my most recent post-doctoral research, I used deep learning technology for solving various problems like- Automatic speech recognition, acoustic and language modeling - Opinion mining, sentiment analysis and text classification - grapheme-to-phoneme (G2P) conversion - Speech emotion recognition My PhD research was focused on large vocabulary continuous speech recognition (LVCSR) for morphologically rich languages like; Arabic, German, and Polish. I investigated diffetent language modeling techniques to handle the special properties of such languages. I did research in the following directions- Open vocabulary ASR - Sub-lexical LMs using different types of sub-lexical units such as morphemes or syllables- Incorporating word/sub-word pronunciations into LMs- Explicit modeling of Arabic diacritization into Arabic LMs - Extended hybrid LMs containing multi-types of units- Investigating the use of feature-rich LMs which incorporate morphological, syntactic, and data driven features into LMs, where features are created on word or morpheme level- Factored Language Models (FLMs)- Class-based Language Models (CLM)- Stream-based Language Models (SLM)- Hierarchical Pitman-Yor Language Models (HPYLM)- Continuous space LMs using Feed-Forward Deep Neural Networks (FFDNNLM). For my publications: check my Google Scholar profile & hl=en