Abstract
Text-to-speech systems increasingly process user-generated text (UGT) such as ppl and imo, whose pronunciation must be inferred from the canonical rather than surface form. We introduce UGTPhon, the first grapheme-to-phoneme (G2P) benchmark for UGT in English, Vietnamese, and Korean, together with an inference-grounded taxonomy for fine-grained diagnosis. Existing G2P models and frontier LLMs exhibit a systematic canonical-to-non-canonical performance gap, reaching up to 66.8 PER points. As a benchmark baseline, we propose a simple compositional G2P approach that incorporates canonical-form evidence through exact-match lookup and staged decoding. Across matched ByT5 and Qwen2.5-0.5B backbones, explicit canonical-form modeling consistently reduces non-canonical G2P errors. The 0.5B variant also performs competitively with much larger few-shot frontier LLMs, highlighting the benefit of explicitly modeling canonical-form inference for UGT phonemization.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Jeon, M., Park, Y., Park, H. S., Kim, J. H., Kim, D. J., & Lee, H. (2026). Phonemizing User-Generated Text: A Benchmark, Taxonomy, and Compositional Approach. https://omanscience.com/en/articles/phonemizing-user-generated-text-a-benchmark-taxonomy-and-compositional-approach
MLA 9
Jeon, MinJu, et al. "Phonemizing User-Generated Text: A Benchmark, Taxonomy, and Compositional Approach." https://omanscience.com/en/articles/phonemizing-user-generated-text-a-benchmark-taxonomy-and-compositional-approach.
Chicago (author–date)
Jeon, MinJu, Younghan Park, Han Sung Park, Jong-Hwan Kim, Dong-Jin Kim, and Hoyeon Lee. 2026. "Phonemizing User-Generated Text: A Benchmark, Taxonomy, and Compositional Approach." https://omanscience.com/en/articles/phonemizing-user-generated-text-a-benchmark-taxonomy-and-compositional-approach.
Harvard
Jeon, M., Park, Y., Park, H. S., Kim, J. H., Kim, D. J. and Lee, H. (2026) 'Phonemizing User-Generated Text: A Benchmark, Taxonomy, and Compositional Approach', Available at: https://omanscience.com/en/articles/phonemizing-user-generated-text-a-benchmark-taxonomy-and-compositional-approach.
Vancouver
Jeon M, Park Y, Park HS, Kim JH, Kim DJ, Lee H. Phonemizing User-Generated Text: A Benchmark, Taxonomy, and Compositional Approach. https://omanscience.com/en/articles/phonemizing-user-generated-text-a-benchmark-taxonomy-and-compositional-approach
IEEE
M. Jeon, Y. Park, H. S. Park, J. H. Kim, D. J. Kim, and H. Lee, "Phonemizing User-Generated Text: A Benchmark, Taxonomy, and Compositional Approach," https://omanscience.com/en/articles/phonemizing-user-generated-text-a-benchmark-taxonomy-and-compositional-approach.