Related Experiment Video
Updated: Jun 13, 2026
06:09
P300-Based Brain-Computer Interface Speller Performance Estimation with Classifier-Based Latency Estimation
Published on: September 8, 2023
A speech rate estimator using hidden markov models - biomed 2010
Monali V Mujumdar1, Robert F Kubichek
1University of Wymoning, Laramie, WY.
Biomedical Sciences Instrumentation
|May 15, 2010
Summary
This study presents a hidden Markov model method to estimate syllables for calculating speaking rate. The approach demonstrated good accuracy on speech datasets, aiding in speech disorder assessment and automatic speech recognition.
Area of Science:
- Speech Science
- Computational Linguistics
- Signal Processing
Background:
- Speaking rate and articulation rate are crucial metrics in speech disorder assessment.
- These rates also impact automatic speech recognition (ASR) system performance, especially at high speech rates.
Purpose of the Study:
- To introduce a novel hidden Markov model (HMM)-based method for estimating the number of syllables in speech.
- To enable more accurate calculation of speaking rate and articulation rate using syllable counts.
Main Methods:
- Utilized a hidden Markov model (HMM) framework for syllable estimation.
- Employed the Viterbi state sequence derived from the HMM for syllable counting.
- Evaluated the method on the TIMIT and Switchboard speech datasets.
Main Results:
- Achieved a 14.6% error rate and 0.81 correlation with TIMIT data.
- Obtained a 21.2% error rate and 0.88 correlation with Switchboard data.
- Demonstrated the efficacy of the Viterbi sequence for syllable estimation.
Conclusions:
- The proposed HMM-based syllable estimation method is effective for calculating speech rates.
- This technique can improve the assessment of speech disorders and enhance ASR system robustness.
- The Viterbi state sequence provides a reliable basis for syllable counting in speech analysis.
