Allophone clustering for continuous speech recognition

Kai Fu Lee*, Satoru Hayamizu, Hsiao Wuen Hon, Cecil Huang, Jonathan Swartz, Robert Weide

*Corresponding author for this work

Research output: Contribution to journalConference articlepeer-review

40 Citations (Scopus)


Two methods are presented for subword clustering. The first method is an agglomerative clustering algorithm. This method is completely data-driven and finds clusters without any external guidance. The second method uses decision trees for clustering. This method uses an expert-generated list of questions about contexts and recursively selects the most appropriate question to split the allophones. Preliminary results showed that when the training set has a good coverage of the allophonic variations in the test set, both methods are capable of high-performance recognition. However, under vocabulary-independent conditions, the method using tree-based allophones outperformed agglomerative clustering because of its superior generalization capability.

Original languageEnglish
Pages (from-to)749-752
Number of pages4
JournalICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Publication statusPublished - 1990
Externally publishedYes
Event1990 International Conference on Acoustics, Speech, and Signal Processing: Speech Processing 2, VLSI, Audio and Electroacoustics Part 2 (of 5) - Albuquerque, New Mexico, USA
Duration: 1990 Apr 31990 Apr 6

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Electrical and Electronic Engineering


Dive into the research topics of 'Allophone clustering for continuous speech recognition'. Together they form a unique fingerprint.

Cite this