Sound event synthesis from onomatopoeic words

We propose two methods of environmental sound synthesis from onomatopoeic words on the basis of the sequence-to-sequence conversion framework as follows:

・Environmental sound synthesis using only onomatopoeic words (seq2seq)

・Environmental sound synthesis using onomatopoeic words and sound event labels (seq2seq w/ event label conditioning)

Demo page can be found here!!

Paper: https://arxiv.org/abs/2102.05872