Lune

ACL Findings2026

Hard to Be Heard: Phoneme-Level ASR Analysis of Phonologically Complex, Low-Resource Endangered Languages

V. S. D. S. Mahesh Akavarapu, Michael Daniel, Gerhard Jäger

Abstract

We present a phoneme-level analysis of automatic speech recognition (ASR) for two lowresourced and phonologically complex East Caucasian languages, Archi and Rutul, based on curated and standardized speech-transcript resources totaling approximately 50 minutes and 1 hour 20 minutes of audio, respectively. Existing recordings and transcriptions are consolidated and processed into a form suitable for ASR training and evaluation. We evaluate several state-of-the-art audio and audiolanguage models, including wav2vec2, Whisper, and Qwen2-Audio. For wav2vec2, we introduce a language-specific phoneme vocabulary with heuristic output-layer initialization, which yields consistent improvements and achieves performance comparable to or exceeding Whisper in these extremely low-resource settings. Beyond standard word and character error rates, we conduct a detailed phonemelevel error analysis. We find that phoneme recognition accuracy strongly correlates with training frequency, exhibiting a characteristic sigmoid-shaped learning curve. For Archi, this relationship partially breaks for Whisper, pointing to model-specific generalization effects beyond what is predicted by training frequency. Overall, our results indicate that many errors attributed to phonological complexity are better explained by data scarcity. These findings demonstrate the value of phoneme-level evaluation for understanding ASR behavior in lowresource, typologically complex languages.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Robust Speech Recognition via Large-Scale Weak Supervision cites wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
Builds onCited byLeft to right by year; the more cited, the higher.

Builds on4

  1. Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
  2. wav2vec 2.0: A Framework for Self-Supervised Learning of Speech RepresentationsAlexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, Michael AuliNeurIPS 2020 · 9,451 citations
  3. Robust Speech Recognition via Large-Scale Weak SupervisionAlec Radford, Jong Wook Kim, Tao Xu, Greg Brockman et al.ICML 2023 · 6,966 citations
  4. Phoneme transcription of endangered languages: an evaluation of recent ASR architectures in the single speaker scenarioGilles BoulianneACL Findings 2022 · 1 citation

Related papers

  1. Beyond WER: Probing Whisper's Sub-token Decoder Across Diverse Language Resource LevelsSiyu Liang, Nicolas Ballier, Gina-Anne Levow, Richard A. WrightEMNLP 2025
  2. Language Complexity and Speech Recognition Accuracy: Orthographic Complexity Hurts, Phonological Complexity Doesn'tChihiro Taguchi, David ChiangACL 2024
  3. Towards Building ASR Systems for the Next Billion UsersTahir Javed, Sumanth Doddapaneni, Abhigyan Raman, Kaushal Santosh Bhogale et al.AAAI 2022 · 86 citations
  4. From Tens of Hours to Tens of Thousands: Scaling Back-Translation for Speech RecognitionTianduo Wang, Lu Xu, Wei Lu, Shanbo ChengEMNLP 2025 · 1 citation
  5. Unsupervised Speech RecognitionAlexei Baevski, Wei-Ning Hsu, Alexis Conneau, Michael AuliNeurIPS 2021 · 309 citations
Dusk over the sea between two cliffs drawn in fine vertical lines