Saved in:
Bibliographic Details
Main Authors: Peckham, Leo, Ong, Michael, Nagy, Naomi, Dunbar, Ewan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2508.11771
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915452426387456
author Peckham, Leo
Ong, Michael
Nagy, Naomi
Dunbar, Ewan
author_facet Peckham, Leo
Ong, Michael
Nagy, Naomi
Dunbar, Ewan
contents We examine the role of transcription inconsistencies in the Faetar Automatic Speech Recognition benchmark, a challenging low-resource ASR benchmark. With the help of a small, hand-constructed lexicon, we conclude that find that, while inconsistencies do exist in the transcriptions, they are not the main challenge in the task. We also demonstrate that bigram word-based language modelling is of no added benefit, but that constraining decoding to a finite lexicon can be beneficial. The task remains extremely difficult.
format Preprint
id arxiv_https___arxiv_org_abs_2508_11771
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Investigating Transcription Normalization in the Faetar ASR Benchmark
Peckham, Leo
Ong, Michael
Nagy, Naomi
Dunbar, Ewan
Computation and Language
We examine the role of transcription inconsistencies in the Faetar Automatic Speech Recognition benchmark, a challenging low-resource ASR benchmark. With the help of a small, hand-constructed lexicon, we conclude that find that, while inconsistencies do exist in the transcriptions, they are not the main challenge in the task. We also demonstrate that bigram word-based language modelling is of no added benefit, but that constraining decoding to a finite lexicon can be beneficial. The task remains extremely difficult.
title Investigating Transcription Normalization in the Faetar ASR Benchmark
topic Computation and Language
url https://arxiv.org/abs/2508.11771