Based On An Audio Match Of The Smart Broadcast Advertisements

Based on traditional broadcast advertising supervisor mode is more difficult to bring the recorder to maintain a large number of videotapes of the store occupied a lot of space, video loss serious, is not conducive to save and find advertising difficulties, shortcomings in the efficiency of low exposure, this paper designs a kinds of automatic, intelligent advertising broadcast monitoring system for television stations and advertisers to provide accurate and credible data reports and a videotape. The system uses high-speed DSP-based high-speed, high-precision speech recognition algorithms, the use of computer information retrieval techniques in real-time automatic analysis of television programs, statistics, calibration-specific advertising content, relevant historical information will be recorded, stored, Avoid listening when the artificial time-consuming, laborious and other defects. At the same time this is also based on the actual project to design a set of suitable advertising regulatory system, which can mainly be divided into video capture, audio, matching and the user interface in these parts, and focuses on the regulation of audio for TV ads matching algorithm.Implementation of radio, cable TV programs, broadcast advertising automatically identify and analyze statistics. In this paper, summing up speech recognition and music retrieval technology based on the use of COFDM channel coding and modulation techniques, the successful implementation of the advertising audio matching algorithm, the algorithm can be divided into the audio feature extraction and template matching audio in two parts.In the audio feature extraction stage, this paper, MPEG-7audio encoding, through pattern recognition, signal processing, music, psychology, cross-disciplinary techniques to identify the sound characteristics.Template matching in the audio stage, the use of AES/EBU audio interface protocols and COFDM coding to solve the audio ads appear in multi-frame, the amount of the phenomenon of dropped frames, while this paper, vector quantization techniques and DTW algorithms.In the vector quantization, the entire ad for the sample database is difficult to establish a unified code of this situation, this paper ad for a single code of this method were established and quantified should not belong to the advertising space or divide a vector into another filter area to improve the match accuracy.Matching algorithm in determining the audio after the experiment proved that the algorithm is simple and strong anti-interference ability, advertising recognition, high accuracy!...