Automatic Heteronym Resolution Pipeline Using RAD-TTS Aligners
This addresses the challenge of limited and expensive human-labeled heteronym datasets for text-to-speech applications, though it appears incremental as it builds on existing aligner technology.
The paper tackles the problem of heteronym disambiguation in grapheme-to-phoneme conversion by proposing an automatic pipeline using RAD-TTS Aligners to generate and score pronunciation candidates from audio-text pairs, enabling the creation of training datasets for G2P systems.
Grapheme-to-phoneme (G2P) transduction is part of the standard text-to-speech (TTS) pipeline. However, G2P conversion is difficult for languages that contain heteronyms -- words that have one spelling but can be pronounced in multiple ways. G2P datasets with annotated heteronyms are limited in size and expensive to create, as human labeling remains the primary method for heteronym disambiguation. We propose a RAD-TTS Aligner-based pipeline to automatically disambiguate heteronyms in datasets that contain both audio with text transcripts. The best pronunciation can be chosen by generating all possible candidates for each heteronym and scoring them with an Aligner model. The resulting labels can be used to create training datasets for use in both multi-stage and end-to-end G2P systems.