Speech2Vec: A Sequence-to-Sequence Framework for Learning Word Embeddings from Speech

TL;DR

Speech2Vec is a neural network model that learns semantic word embeddings directly from speech audio, capturing meaning beyond text-based methods, and outperforms Word2Vec on multiple benchmarks.

Contribution

It introduces a speech-based word embedding model using an RNN Encoder-Decoder architecture, enabling semantic learning directly from speech signals.

Findings

Outperforms Word2Vec on 13 word similarity benchmarks

Learns meaningful semantic representations directly from speech

Demonstrates the potential of speech-based embeddings for NLP tasks

Abstract

In this paper, we propose a novel deep neural network architecture, Speech2Vec, for learning fixed-length vector representations of audio segments excised from a speech corpus, where the vectors contain semantic information pertaining to the underlying spoken words, and are close to other vectors in the embedding space if their corresponding underlying spoken words are semantically similar. The proposed model can be viewed as a speech version of Word2Vec. Its design is based on a RNN Encoder-Decoder framework, and borrows the methodology of skipgrams or continuous bag-of-words for training. Learning word embeddings directly from speech enables Speech2Vec to make use of the semantic information carried by speech that does not exist in plain text. The learned word embeddings are evaluated and analyzed on 13 widely used word similarity benchmarks, and outperform word embeddings learned by Word2Vec from the transcriptions.