# Tone on a Budget: A Reference-Free Metric for Lexical Tone in Massively Multilingual Text-to-Speech

arXiv: https://arxiv.org/abs/2609.14817 · HTML: https://mlsanity.com/paper/2609.14817/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 12 of its 29 references that are in the map.
- First posted on arXiv 13 Sep 2026.

## Builds on (most cited first, 12 of 12)

- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Scaling Speech Technology to 1,000+ Languages](https://mlsanity.com/paper/2305.13516/) (2023)
- [BibleTTS: a large, high-fidelity, multilingual, and uniquely African speech corpus](https://mlsanity.com/paper/2207.03546/) (2022)
- [EmergentTTS-Eval: Evaluating TTS Models on Complex Prosodic, Expressiveness, and Linguistic Challenges Using Model-as-a-Judge](https://mlsanity.com/paper/2505.23009/) (2025)
- [Prosodic ABX: A Language-Agnostic Method for Measuring Prosodic Contrast in Speech Representations](https://mlsanity.com/paper/2604.02102/) (2026)
- [Objective Evaluation of Prosody and Intelligibility in Speech Synthesis via Conditional Prediction of Discrete Tokens](https://mlsanity.com/paper/2509.20485/) (2025)
- [Linguistically Informed Evaluation of Multilingual ASR for African Languages](https://mlsanity.com/paper/2602.04716/) (2026)
- [LTA-L2S: Lexical Tone-Aware Lip-to-Speech Synthesis for Mandarin with Cross-Lingual Transfer Learning](https://mlsanity.com/paper/2509.25670/) (2025)
- [Vox-Evaluator: Enhancing Stability and Fidelity for Zero-shot TTS with A Multi-Level Evaluator](https://mlsanity.com/paper/2510.20210/) (2025)
- [Towards Prosodically Informed Mizo TTS without Explicit Tone Markings](https://mlsanity.com/paper/2601.02073/) (2026)
- [Lexical Tone is Hard to Quantize: Probing Discrete Speech Units in Mandarin and Yor`ub'a](https://mlsanity.com/paper/2604.07467/) (2026)
- [Towards Digital Preservation of Efik: TTS for a Low-Resource African Language](https://mlsanity.com/paper/2607.04515/) (2026)

## Abstract (arXiv)

In Yor`ub'a, pitch alone separates dokdo (husband, Mid), dokd`o (vehicle, Low), and dokd'o (hoe, High) -- the diacritics ARE the tone marks. Yet character error rate (CER), the standard automated metric for text-to-speech (TTS), is in practice computed from ASR output that drops those marks: a synthesizer can ace CER and still say vehicle for husband. We introduce DunDun -- named for the d`und'un, the Yor`ub'a talking drum that speaks through pitch alone -- an automated, reference-free lexical-tone metric that needs no tone-labelled corpus. The gold High/Mid/Low sequence is read from the input text's diacritics (in TTS that text exists by construction, so no reference recording is needed); the prediction comes from the audio's pitch track. We validate three ways. Flattening pitch with PSOLA resynthesis collapses DunDun while CER does not move. Inverting High and Low in the answer key of 300 native recordings drives the two-class readout to 0.14, symmetrically below its 0.35 chance level -- a consistency check on the scoring path, not independent evidence. And three native listeners, over 67 blind A/B trials, pick the tone-correct clip 89.6% of the time (95% CI 80.0-94.8; p < 1e-4); whether DunDun tracks those judgements trial by trial is not resolved at this sample size. Applied to a massively multilingual zero-shot TTS model, DunDun shows what CER cannot: Yor`ub'a tone sits near the native anchor before any Yor`ub'a fine-tuning (0.567 +/- 0.02 over five decode seeds vs. 0.596; chance 0.33), despite the 21.4% CER the model's own paper reports; and a few hours of clean audio halve CER (5.6% to 2.7% by 5h, 1.7% by 15h) while tone saturates within the hour. On non-tonal Swahili, CER already captures the gains: the metric a language needs is language-dependent. We release the metric and the complete validation protocol.