An embedded segmental K-means model for unsupervised segmentation and clustering of speech

Herman Kamper, Karen Livescu, Sharon Goldwater

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Unsupervised segmentation and clustering of unlabelled speech are core problems in zero-resource speech processing. Most approaches lie at methodological extremes: some use probabilistic Bayesian models with convergence guarantees, while others opt for more efficient heuristic techniques. Despite competitive performance in previous work, the full Bayesian approach is difficult to scale to large speech corpora. We introduce an approximation to a recent Bayesian model that still has a clear objective function but improves efficiency by using hard clustering and segmentation rather than full Bayesian inference. Like its Bayesian counterpart, this embedded segmental K-means model (ES-KMeans) represents arbitrary-length word segments as fixed-dimensional acoustic word embeddings. We first compare ES-KMeans to previous approaches on common English and Xitsonga data sets (5 and 2.5 hours of speech): ES-KMeans outperforms a leading heuristic method in word segmentation, giving similar scores to the Bayesian model while being 5 times faster with fewer hyperparameters. However, its clusters are less pure than those of the other models. We then show that ES-KMeans scales to larger corpora by applying it to the 5 languages of the Zero Resource Speech Challenge 2017 (up to 45 hours), where it performs competitively compared to the challenge baseline.
Original languageEnglish
Title of host publication2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Pages719-726
Number of pages8
ISBN (Electronic)978-1-5090-4788-8, 978-1-5090-4787-1
ISBN (Print)978-1-5090-4789-5
DOIs
Publication statusPublished - 25 Jan 2018
Event2017 IEEE Automatic Speech Recognition and Understanding Workshop - Okinawa, Japan
Duration: 16 Dec 201720 Dec 2017
https://asru2017.org/

Conference

Conference2017 IEEE Automatic Speech Recognition and Understanding Workshop
Abbreviated titleASRU 2017
CountryJapan
CityOkinawa
Period16/12/1720/12/17
Internet address

Keywords

  • Bayes methods
  • linguistics
  • natural language processing
  • pattern clustering
  • speech processing
  • speech recognition
  • unsupervised learning
  • competitive performance
  • Bayesian approach
  • speech corpora
  • hard clustering
  • Bayesian inference
  • fixed-dimensional acoustic word embeddings
  • word segmentation
  • ES-KMeans scales
  • zero-resource speech processing
  • convergence guarantees
  • heuristic techniques
  • probabilistic Bayesian models
  • Xitsonga data sets
  • English data sets
  • time 5.0 hour
  • time 2.5 hour
  • time 45.0 hour
  • Speech
  • Clustering algorithms
  • Acoustics
  • Standards
  • Speech processing
  • Probabilistic logic
  • Zero-resource speech processing
  • language acquisition

Fingerprint Dive into the research topics of 'An embedded segmental K-means model for unsupervised segmentation and clustering of speech'. Together they form a unique fingerprint.

Cite this