Method for recognizing audio based on spectrogram significance test
An audio recognition and spectrogram technology, applied in speech analysis, instruments, etc., can solve the problem of unpractical audio data
Image
Examples
Embodiment Construction
[0045] In this example, if figure 1 As shown, an audio recognition method based on spectrogram saliency detection is performed in the following steps:
[0046] Step 1. Obtain the spectrograms of n different sound sources with M×N pixels, and obtain m pieces of each kind of spectrogram, so as to obtain m×n spectrograms D={d 1 , d 2 ,...,d i ,...,d m×n}; d i Indicates the i-th spectrogram; i∈[1,m×n];
[0047] Carry out feature extraction respectively to m×n spectrogram D, obtain basic feature set; Basic feature set includes: RGBY chromaticity feature set C={C 1 ,C 2 ,...,C i ,...,C m×n}, direction feature set O={O 1 ,O 2 ,...,O i ,...,O m×n} and brightness feature set I={I 1 , I 2 ,...,I i ,...,I m×n};C i Indicates the i-th spectrogram d i RGBY chromaticity characteristics; O i Indicates the i-th spectrogram d i directional characteristics; I i Indicates the i-th spectrogram d i The brightness feature; in the present embodiment, the direction feature is rep...
PUM
Login to View More Abstract
Description
Claims
Application Information
- IPC
- G10L25/03; G10L25/48
- Inventors
- 陈雁翔; 弓彦婷