Pitch Period Estimation for Given Speech Corpus
- Login to Download
- 1 Credits
Resource Overview
1. Implement pitch detection algorithms in MATLAB to estimate pitch periods for the given speech corpus, providing complete detection results and analysis. Key implementation should include algorithms like autocorrelation, cepstrum analysis, or AMDF (Average Magnitude Difference Function) with appropriate parameter configurations. 2. Experimental corpus: aoYiSheng.raw containing pronunciation "Mao Yisheng"; duration 0.9 seconds; sampling rate: 8kHz; quantization precision: 16 bits per sample; data format: LSB, MSB; analysis frame length: 20ms (160 samples); total frames: 45
Detailed Documentation
1. Implement pitch detection algorithms using MATLAB programming language to estimate pitch periods for the specified speech corpus. The implementation should include robust pitch tracking methods with proper pre-processing steps such as framing, windowing (Hamming window recommended), and endpoint detection. Provide complete detection results including pitch contour visualization and statistical analysis. 2. The experimental corpus aoYiSheng.raw contains the pronunciation "Mao Yisheng" with the following specifications: duration 0.9 seconds, sampling rate 8kHz, quantization precision 16 bits per sample, data format LSB and MSB. The analysis uses 20ms frame length (160 samples per frame) resulting in 45 total frames. The implementation should handle raw audio file reading, byte order conversion for LSB/MSB formats, and include error handling for invalid frames. Key MATLAB functions to utilize include audioread (with appropriate parameters for raw files), xcorr for autocorrelation, and fft for frequency domain analysis. Consider implementing multiple pitch detection methods for comparative analysis and validation.
- Login to Download
- 1 Credits