Lightly supervised recognition for automatic alignment of large coherent speech recordings

115 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Unsupervised clustering of emotion and voice styles for expressive TTS (2012, methodology): “To be confident of ensuring no mismatch between the text and audio, only units for which the lightly supervised segmentation [4] achieved 100% word accuracy against the book text were retained for training.”
Audio-to-text alignment for speech recognition with very limited resources (2014, background): “Other ways to train speech recognition systems with low resources include the adaptation of a preexisting ASR system on the same language, but different dialect [10], cross-language bootstrapping [15] or training seed acoustic models from very little manually aligned data [16].”
Expressive Speech Synthesis via Modeling Expressions with Variational Autoencoder (2018, methodology): “Our second dataset was from the Blizzard Challenge 2012 (Blizzard2012), and consisting of four audiobooks [27, 28].”