CISPA
Browse
17_afrinames_most_asr_models_butc.pdf (313.17 kB)

AfriNames: Most ASR models "butcher" African Names.

Download (313.17 kB)
conference contribution
posted on 2024-05-24, 07:54 authored by Tobi Olatunji, Tejumade AfonjaTejumade Afonja, Bonaventure FP Dossou, Atnafu Lambebo Tonja, Chris Chinenye Emezue, Amina Mardiyyah Rufai, Sahib Singh
Useful conversational agents must accurately capture named entities to minimize error for downstream tasks, for example, asking a voice assistant to play a track from a certain artist, initiating navigation to a specific location, or documenting a diagnosis result for a specific patient. However, where named entities such as "Ukachukwu" (Igbo), "Lakicia" (Swahili), or "Ingabire" (Rwandan) are spoken, automatic speech recognition (ASR) models' performance degrades significantly, propagating errors to downstream systems. We model this problem as a distribution shift and demonstrate that such model bias can be mitigated through multilingual pre-training, intelligent data augmentation strategies to increase the representation of African named entities, and fine-tuning multilingual ASR models on multiple African accents. The resulting fine-tuned models show an 86.4% relative improvement compared with the baseline on samples with African named entities.

History

Primary Research Area

  • Trustworthy Information Processing

Name of Conference

International Conference on Learning Representations (ICLR)

Journal

AfricaNLP

BibTeX

@conference{Olatunji:Afonja:Dossou:Tonja:Emezue:Rufai:Singh:2023, title = {AfriNames: Most ASR models "butcher" African Names.}, author = "Olatunji, Tobi" AND "Afonja, Tejumade" AND "Dossou, Bonaventure FP" AND "Tonja, Atnafu Lambebo" AND "Emezue, Chris Chinenye" AND "Rufai, Amina Mardiyyah" AND "Singh, Sahib", year = 2023, month = 3, journal = "AfricaNLP" }

Usage metrics

    Categories

    No categories selected

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC