Spoken Word2Vec: Learning Skipgram Embeddings from Speech

Mohammad Amaan Sayeed, Hanan Aldarmaki
11/15/2023
cs.CLcs.AI

No official code found. Best implementation: rainmaker29/SpokenWord2Vec (4 stars, last commit 2024-08-31).

Abstract

Text word embeddings that encode distributional semantics work by modeling contextual similarities of frequently occurring words. Acoustic word embeddings, on the other hand, typically encode low-level phonetic similarities. Semantic embeddings for spoken words have been previously explored using analogous algorithms to Word2Vec, but the resulting vectors still mainly encoded phonetic rather than semantic features. In this paper, we examine the assumptions and architectures used in previous works and show experimentally how shallow skipgram-like algorithms fail to encode distributional semantics when the input units are acoustically correlated. We illustrate the potential of an alternative deep end-to-end variant of the model and examine the effects on the resulting embeddings, showing positive results of semantic relatedness in the embedding space.

View on arXivView PDF

Code Implementations(1)

rainmaker29/SpokenWord2VecCommunity100%

Implementation of the paper "" . We're in progress of adding code and data files to the repository.

42PythonNov 15, 20232 years ago

Similar Papers

Mimicking Word Embeddings using Subword RNNs

81% match1 repo1532017

Construction of a Japanese Word Similarity Dataset

80% match1 repo1032017

A Theoretical Framework for Acoustic Neighbor Embeddings

80% match2024

Contrastive Loss is All You Need to Recover Analogies as Parallel Lines

80% match2023

PWESuite: Phonetic Word Embeddings and Tasks They Facilitate

80% match1 repo382023

On the Sentence Embeddings from Pre-trained Language Models

80% match1 repo5342020

Cite this paper

@article{sayeed2023spoken,
  title  = {Spoken Word2Vec: Learning Skipgram Embeddings from Speech},
  author = {Mohammad Amaan Sayeed and Hanan Aldarmaki},
  year   = {2023},
  eprint = {2311.09319},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2311.09319v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2311.09319.svg)](https://codeofpaper.com/paper/2311.09319)

Discussion