Abner Hernandez

Computational Linguist | Speech Scientist | NLP Engineer

Barcelona, Catalonia, Spain

About

Hello! I'm a computational linguist with experience in speech recognition, NLP, and machine learning applications to linguistics and speech. I have a mix of industry and research experience.

Experience

  • Speech & NLP Researcher at Pattern Recognition Lab
    May 2025 - Present · 1 yr 2 mos

  • Speech Recognition Engineer at Cerence Inc.
    Jul 2023 - Aug 2024 · 1 yr 2 mos

    - Develop and train acoustic models for multilingual speech recognition systems. - Managed training data collection and processing for both new and legacy acoustic models. - Implemented data augmentation methods to improve model performance in noisy conditions. - Evaluated ASR systems using metrics to assess performance.

  • Language Engineer at Apple
    May 2022 - Jul 2023 · 1 yr 3 mos

    - Analyze speech and language errors in Siri across various English dialects. - Create, review, and refine synthetic datasets and ontologies for dialogue schemas. - Develop, verify and correct annotations for semantic schemas and phonetic lexicons. - Collaborate with cross-functional teams to enhance support for European and Asian languages. - Diagnose bugs, propose solutions, and delegate tasks to linguists and data curators.

  • Speech & NLP Researcher at Pattern Recognition Lab
    Apr 2021 - Apr 2022 · 1 yr 1 mo

    Research Engineer for the Pattern Recognition Lab - Collaborate with the AI for Biomedical Engineering Dept. to enhance dysarthric speech recognition using transformer-based acoustic models. Support disease detection/diagnosis through speech analysis. - Experiment with the utility of self-supervised speech representations for atypical speech processing using multilingual data.

  • Speech Recognition Engineer at Saltlux (솔트룩스)
    Nov 2020 - Feb 2021 · 4 mos

    - Develop an end-to-end speech recognition system for the Korean language. - Evaluate model performance on various deep learning-based acoustic models, N-gram/transformer language models and data augmentation methods. - Research and implement neural network models for Korean, Japanese, and Chinese ASR, with a focus on vocabulary, language models, and data preprocessing.