{"ID":5675070,"CreatedAt":"2026-07-03T01:40:09.565152011Z","UpdatedAt":"2026-07-05T01:41:04.687834994Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.01594","arxiv_id":"2607.01594","title":"Enhancing Acoustic-to-Articulatory Inversion with Multi-Target Pretraining for Low-Resource Settings","abstract":"Acoustic-to-Articulatory Inversion (AAI) estimates vocal tract articulator movements from speech, benefiting tasks like ASR, speech synthesis, and speaker verification. While deep learning-based methods (CNNs, RNNs, Transformers) have advanced AAI, recent studies show that Self-Supervised Learning (SSL) features further enhance performance, particularly in low-resource settings. However, SSL feature extractors introduce inference latency and computational overhead. To address this, we propose a novel pretraining method leveraging three target representations-Phoneme Labels, Articulatory Feature Labels, and Critical-articulator Labels-eliminating the need for an SSL extractor during inference. We evaluate our approach against both baseline and SSL-based models across various data conditions. Results demonstrate that our method consistently improves AAI performance, particularly in low-resource scenarios, while significantly reducing inference costs without sacrificing accuracy.","short_abstract":"Acoustic-to-Articulatory Inversion (AAI) estimates vocal tract articulator movements from speech, benefiting tasks like ASR, speech synthesis, and speaker verification. While deep learning-based methods (CNNs, RNNs, Transformers) have advanced AAI, recent studies show that Self-Supervised Learning (SSL) features furthe...","url_abs":"https://arxiv.org/abs/2607.01594","url_pdf":"https://arxiv.org/pdf/2607.01594v1","authors":"[\"Jesuraj Bandekar\",\"Prasanta Kumar Ghosh\"]","published":"2026-07-02T01:43:57Z","proceeding":"eess.AS","tasks":"[\"eess.AS\"]","methods":"[\"Transformer\",\"Convolutional Neural Network\"]","has_code":false}