Unsupervised Word Segmentation from Speech with Attention

We present a first attempt to perform attentional word segmentation from speech signal, with the final goal of automatically identifying lexical units in a low-resource, unwritten language (UL). Our methodology assumes a pairing between recordings in the UL with translations in a well-resourced language. It uses Acoustic Unit Discovery (AUD) to convert speech into a pseudo-phones sequence that is segmented using neural soft alignments (from a neural machine translation model). Evaluation uses an actual Bantu UL, Mboshi; comparisons to monolingual and bilingual baselines illustrate the potential of attentional word segmentation for language documentation.