Skip to main content
. 2024 Mar 5;18:1360094. doi: 10.3389/fnbot.2024.1360094

Table 2.

The impact of accents on different speech recognition models.

Model classification Model characteristics Models composition Impacts
Traditional speech recognition models Training and optimizing the acoustic model and language model separately Acoustical model Changes in acoustic features
Language model Lead to changes in vocabulary usage and order, making it more challenging for language models to understand.
End-to-end speech recognition models From acoustic features to integrated text Encoder-decoder The degradation of speech recognition generalization and adaptability