Font Size: a A A

A Novel Audience Rating System And Search Method Based On Audio Features

Posted on:2012-11-24Degree:MasterType:Thesis
Country:ChinaCandidate:X X WangFull Text:PDF
GTID:2218330338462171Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Audience rating survey equipment is for investigating television viewing after questionnaire. It has been widely used in broadcast department. And it has high study value. The result plays an important part for TV station arranging program, advertisement and audience. At present, the principle of the devices is diversity and the cost of equipment is higher. The reliability needs improvement. Under this circumstance, we have studied a new type of audience rating survey equipment based on audio feature and a new audio retrieval method.First, we introduce some knowledge of audience rating and survey, such as indexes, methods and significance. Then talk about the current audience rating survey devices and their fault. Last, we propose the research content:the system of audience rating survey based audio feature, and audio signal dispose methods and audio retrieval technologies are also illustrated.System of audience rating survey based audio feature consists of data capture and dispose parts. Besides, the communication module is also included. We introduce the detail function and the problem of the three parts and the solution.This thesis commit to designing feature vector and a new audio retrieval method for audience rating survey equipment based audio for cost reduction and device reliability. The designed feature improves the factual problem, uses passband division technique and eliminates sensitive sound passband. Meanwhile, normalization method avoids the influence of the various volumes to the audio feature. Last, we analyses and study the reliability through chart on MATLAB platform. Audio retrieval method includes two steps. First, correlation computed between under test audio and other audio in the library. Second, compare the Euclidean distance and adopt multi-packet data discrimination method to ensure accuracy of the result. Last, we make evaluation the method on MATLAB platform. Besides, the data features are all on the basis of our constructed system and satisfy small number and high validity. Therefore, the device and data processing complexity and communication cost of PRI are all reduced. The communication module has three interfaces. Thus, it can satisfy telephone, wireless and network transfer. Lastly, we make a summary and point out some improvement.
Keywords/Search Tags:audience rating survey, audio feature, audio retrieval method
PDF Full Text Request
Related items