Proposes a classifier framework that audits TTS output for phonological faithfulness using human speech benchmarks, revealing realization biases in Meta's MMS TTS for Assamese ATR harmony.
Data and model setup Human benchmark.We created the human benchmark corpus by recording the speech of14adult native Assamese speakers from the upper Assam region (8females,6males)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Towards a Phonology-Informed Evaluation of Multilingual TTS
Proposes a classifier framework that audits TTS output for phonological faithfulness using human speech benchmarks, revealing realization biases in Meta's MMS TTS for Assamese ATR harmony.