2021/04/11 by Sathvik Udupa, Anwesha Roy, Udupa, Sathvik +7
Computer Science · Psychology · #Audio and Speech Processing (eess.AS) #FOS: Computer and information sciences #FOS: Electrical engineering #Phonetics and Phonology Research #Sound (cs.SD) #Speech Recognition and Synthesis #Speech and Audio Processing #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2104.05017
openalex publication_date 2021/04/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We estimate articulatory movements in speech production from different\nmodalities - acoustics and phonemes. Acoustic-to articulatory inversion (AAI)\nis a sequence-to-sequence task. On the other hand, phoneme to articulatory\n(PTA) motion estimation faces a key challenge in reliably aligning the text and\nthe articulatory movements. To address this challenge, we explore the use of a\ntransformer architecture - FastSpeech, with explicit duration modelling to\nlearn hard alignments between the phonemes and articulatory movements. We also\ntrain a transformer model on AAI. We use correlation coefficient (CC) and root\nmean squared error (rMSE) to assess the estimation performance in comparison to\nexisting methods on both tasks. We observe 154%, 11.8% & 4.8% relative\nimprovement in CC with subject-dependent, pooled and fine-tuning strategies,\nrespectively, for PTA estimation. Additionally, on the AAI task, we obtain\n1.5%, 3% and 3.1% relative gain in CC on the same setups compared to the\nstate-of-the-art baseline. We further present the computational benefits of\nhaving transformer architecture as representation blocks.\n