# Scaling Human and G2P Supervision for Robust Phonetic Transcription

arXiv: https://arxiv.org/abs/2606.16019 · HTML: https://mlsanity.com/paper/2606.16019/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 49 references that are in the map.
- First posted on arXiv 14 Jun 2026.

## Builds on (most cited first, 22 of 22)

- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://mlsanity.com/paper/2006.11477/) (2020)
- [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://mlsanity.com/paper/2106.07447/) (2021)
- [A Survey on Bias and Fairness in Machine Learning](https://mlsanity.com/paper/1908.09635/) (2019)
- [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://mlsanity.com/paper/2110.13900/) (2021)
- [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://mlsanity.com/paper/2111.09296/) (2021)
- [Unsupervised Cross-lingual Representation Learning for Speech Recognition](https://mlsanity.com/paper/2006.13979/) (2020)
- [Universal Phone Recognition with a Multilingual Allophone System](https://mlsanity.com/paper/2002.11800/) (2020)
- [Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://mlsanity.com/paper/2109.11680/) (2021)
- [speechocean762: An Open-Source Non-native English Speech Corpus For Pronunciation Assessment](https://mlsanity.com/paper/2104.01378/) (2021)
- [Transformer-Based Multi-Aspect Multi-Granularity Non-Native English Speaker Pronunciation Assessment](https://mlsanity.com/paper/2205.03432/) (2022)
- [Transformer based Grapheme-to-Phoneme Conversion](https://mlsanity.com/paper/2004.06338/) (2020)
- [Whistle: Data-Efficient Multilingual and Crosslingual Speech Recognition via Weakly Phonetic Supervision](https://mlsanity.com/paper/2406.02166/) (2024)
- [Universal Automatic Phonetic Transcription into the International Phonetic Alphabet](https://mlsanity.com/paper/2308.03917/) (2023)
- [POWSM: A Phonetic Open Whisper-Style Speech Foundation Model](https://mlsanity.com/paper/2510.24992/) (2025)
- [Phoneme Hallucinator: One-shot Voice Conversion via Set Expansion](https://mlsanity.com/paper/2308.06382/) (2023)
- [MultiPA: A Multi-task Speech Pronunciation Assessment Model for Open Response Scenarios](https://mlsanity.com/paper/2308.12490/) (2023)
- [Improving Spoken Language Modeling with Phoneme Classification: A Simple Fine-tuning Approach](https://mlsanity.com/paper/2410.00025/) (2024)
- [HuPER: A Human-Inspired Framework for Phonetic Perception](https://mlsanity.com/paper/2602.01634/) (2026)
- [Phoneme-based Distribution Regularization for Speech Enhancement](https://mlsanity.com/paper/2104.03759/) (2021)
- [Language-universal phonetic encoder for low-resource speech recognition](https://mlsanity.com/paper/2305.11576/) (2023)
- [Linguistic and Audio Embedding-Based Machine Learning for Alzheimer's Dementia and Mild Cognitive Impairment Detection: Insights from the PROCESS Challenge](https://mlsanity.com/paper/2510.03336/) (2025)

## Built on by (most cited first, 1 of 1)

- [Phone Segmentation and Recognition through Phonological Activation Mapping](https://mlsanity.com/paper/2607.09020/) (2026)

## Abstract (arXiv)

Expert phonetic annotation is costly, especially for non-standard dialects and atypical speech. A common alternative is using Grapheme-to-Phoneme (G2P) models to auto-generate phonetic labels from text transcripts at scale. We study how automatic phonetic transcription performance scales with human and G2P supervision in English. Using a curated 80-hour benchmark spanning native, non-native and post-stroke speech, we identify a supervision quality threshold: G2P supervision helps only when fewer than 20-30 hours of human annotation are available. Beyond this threshold, it provides no significant benefit and can reduce cross-dialect robustness. What is effective after this threshold is ASR pretraining which we use to achieve a 2.3x reduction in weighted phone feature error rate over prior systems, with strong gains on non-native and aphasic speech. These results suggest that quantity-driven G2P scaling may yield diminishing returns for robust generalization.