# Phonemizing User-Generated Text: A Benchmark, Taxonomy, and Compositional Approach

arXiv: https://arxiv.org/abs/2609.27205 · HTML: https://mlsanity.com/paper/2609.27205/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 34 references that are in the map.
- First posted on arXiv 23 Sep 2026.

## Builds on (most cited first, 13 of 13)

- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions](https://mlsanity.com/paper/1712.05884/) (2017)
- [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://mlsanity.com/paper/2106.06103/) (2021)
- [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://mlsanity.com/paper/2105.13626/) (2021)
- [ByT5 model for massively multilingual grapheme-to-phoneme conversion](https://mlsanity.com/paper/2204.03067/) (2022)
- [What to do about non-standard (or non-canonical) language in NLP](https://mlsanity.com/paper/1608.07836/) (2016)
- [Transformer based Grapheme-to-Phoneme Conversion](https://mlsanity.com/paper/2004.06338/) (2020)
- [r-G2P: Evaluating and Enhancing Robustness of Grapheme to Phoneme Conversion by Controlled noise introducing and Contextual information incorporation](https://mlsanity.com/paper/2202.11194/) (2022)
- [Neural Grapheme-to-Phoneme Conversion with Pre-trained Grapheme Models](https://mlsanity.com/paper/2201.10716/) (2022)
- [KoMultiText: Large-Scale Korean Text Dataset for Classifying Biased Speech in Real-World Online Services](https://mlsanity.com/paper/2310.04313/) (2023)
- [PolyNorm: Few-Shot LLM-Based Text Normalization for Text-to-Speech](https://mlsanity.com/paper/2511.03080/) (2025)
- [Mitigating the Exposure Bias in Sentence-Level Grapheme-to-Phoneme (G2P) Transduction](https://mlsanity.com/paper/2308.08442/) (2023)
- [MultiLexNorm++: A Unified Benchmark and a Generative Model for Lexical Normalization for Asian Languages](https://mlsanity.com/paper/2601.16623/) (2026)

## Abstract (arXiv)

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.