Weakly-supervised text-to-speech alignment confidence measure

Guillaume Serrière (1) , Christophe Cerisara (1) , Dominique Fohr (2) , Odile Mella (2)

1 SYNALP - Natural Language Processing : representations, inference and semantics
2 MULTISPEECH - Speech Modeling for Facilitating Oral-Based Communication

This work proposes a new confidence measure for evaluating text-to-speech alignment systems outputs, which is a key component for many applications, such as semi-automatic corpus anonymization, lips syncing, film dubbing, corpus preparation for speech synthesis and speech recognition acoustic models training. This confidence measure exploits deep neural networks that are trained on large corpora without direct supervision. It is evaluated on an open-source spontaneous speech corpus and outperforms a confidence score derived from a state-of-the-art text-to-speech aligner. We further show that this confidence measure can be used to fine-tune the output of this aligner and improve the quality of the resulting alignment.

Domaines: Informatique et langage [cs.CL]

Guillaume Serrière, Christophe Cerisara, Dominique Fohr, Odile Mella. Weakly-supervised text-to-speech alignment confidence measure. International Conference on Computational Linguistics (COLING), Dec 2016, Osaka, Japan. ⟨hal-01378355⟩