Significant points: pitch period detection as a problem of segmentation

Area of Science:

• Speech Processing
• Signal Analysis
• Acoustic Phonetics

Background:

• Pitch period detection is crucial for speech analysis.
• Existing methods may not fully capture the complex nature of pitch periods.
• Understanding the segmentation of speech signals is key.

Purpose of the Study:

• To investigate pitch period detection as a segmentation problem.
• To introduce and evaluate a novel algorithm for identifying significant points in speech signals.
• To explore the relationship between visually and auditorily recognized speech segments.

Main Methods:

• Framing pitch period detection as a segmentation task.
• Developing a 'skeletization' algorithm to identify 'significant points'.
• Analyzing the characteristics and significance of the selected points.

Main Results:

• The proposed skeletization algorithm effectively selects significant points.
• These points serve as boundaries for segmenting the speech signal.
• The identified segments show potential for relating visual and auditory speech recognition.

Conclusions:

• Pitch period detection can be successfully modeled as a segmentation problem.
• The skeletization algorithm offers a new approach to identifying critical points in speech.
• Further research can explore the auditory-visual link in speech signal segmentation.