A voiceprint recognition method for buried drainage pipelines based on bse and gmm-hmm
A technology of voiceprint recognition and drainage pipes, which is applied in processing detection response signals, using sound waves/ultrasonic waves/infrasonic waves for material analysis, voice analysis, etc., and can solve problems such as low recognition rate of new working conditions and insufficient working condition types. Achieve the effect of high recognition accuracy and easy operation
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0029] Embodiment 1: The voiceprint recognition method of buried drainage pipeline based on BSE and GMM-HMM, the sound pressure signal data acquisition device is as attached figure 1 shown. The implementation process is as follows: the experimenter controls the sound card by operating the computer equipped with WinLS software to generate a sine sweep sound wave signal with a frequency of 100-6000HZ and a length of 10s. The sound card model is ASUS Xonar Essence STXII 7.1, and the power amplifier model is DRV603PWR. The power is amplified, and then through the telescopic rod, the speaker fixed on the telescopic rod is extended into the pipe to emit sound wave signals. The speaker is EVUM30 underwater speaker, and the model fixed on another telescopic rod is RHSM-10 standard hydrophone The receiver receives the sound wave signal reflected from the pipeline, filters it with a Kemo VBF 40 filter, and digitally uploads it to the computer at a sampling rate of 44100Hz to store the s...
Embodiment 2
[0036] Example 2: In order to verify the robustness of the method proposed by the present invention, we verify the proposed method in a more complex working condition, which is specifically that there are two tee parts in the pipeline and the positions are arbitrary, and three different heights are used at the same time. The stone baffles of 20mm, 40mm and 55mm in height were used to simulate the blockage and were arbitrarily positioned. The way of data collection is as in Example 1. The obtained sound pressure signal is first de-noised by wavelet threshold, the number of decomposition layers is 3, the wavelet basis function is db9, the threshold value is determined by the Stein unbiased risk estimation principle method, and the low-frequency coefficient and the processed high-frequency coefficient are used to reconstruct the signal to obtain the reduced noise. noised signal. Then use the sub-band spectral entropy as a discriminant parameter to perform endpoint detection and ...
Embodiment 3
[0037] Example 3: In order to verify the validity and reliability of the method proposed by the present invention, the specific number, type, location and data collection method of the blockage are as in Example 1, but the water level in the pipeline is different. This embodiment will verify the influence of three different water levels of 0 mm, 10 mm and 20 mm in the pipeline on the validity and reliability of the method proposed by the present invention. First, use the sound pressure signal collected in an anhydrous state to perform wavelet threshold noise reduction, and then use the sub-band spectral entropy as a discriminant parameter to perform endpoint detection and signal segmentation on the noise-reduced signal. The individual sound pressure signal of the blockage, the individual sound pressure signal of the tee and the individual sound pressure signal of the tail end. Then use the sound pressure signal collected in the water state to perform noise reduction, endpoint ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


