Radio Station Logo Identification Technology Research And Engineering Realization

Radio broadcasting is a fundamental strategic resource and an important factor of production for information and communication,transportation,national defence and military and socio-economic development.This paper presents a detailed analysis of the main tasks and characteristics of broadcast call sign in the actual engineering of radio monitoring in China,studies the existing technologies of broadcast call sign recognition using audio retrieval and speech recognition,and analyses the problems and shortcomings of the traditional manual monitoring of broadcast call sign and the existing automated broadcast call sign recognition technologies.In order to achieve the purpose of learning and recognising broadcast call sign and to solve the problem of reduced recognition accuracy due to inconsistent transmission links and noise interference,a broadcast call sign recognition algorithm with adaptive broadcast call sign learning capability is proposed in this paper.To address the shortcomings of the existing audio retrieval-based broadcast call sign recognition technology,which requires manual addition of call sign libraries,poor algorithm robustness and low recognition efficiency,the algorithm uses the cepstrum to learn and recognise broadcast call signs with the sensitive recognition ability of the medium distance components of the logarithmic spectrum after the mixing of two broadcast signals containing call sign information;the signals of multiple real broadcast programmes at the whole timing are used as audio sources for real-time acquisition.The audio segments that are highly similar to each other and the starting time of the segments occurring in the same broadcast programme are analysed through the cepstrum,which is the automatic learning result of the broadcast call sign,and the learning result is put into the station call sign library;the similarity between the broadcast call sign samples and all the call signs in the call sign library is analysed through the cepstrum successively,and the call sign corresponding to the highest similarity is calculated as the recognition result of the broadcast call sign samples.The performance of the proposed algorithm was tested experimentally,and the results proved that the proposed algorithm shows great noise immunity under the conditions of additive Gaussian white noise with different SNRs in recognition of broadcast call sign,especially at low SNRs(SNR less than or equal to 5dB),it can still achieve 95.88% identification accuracy,and the identification efficiency is more in line with the requirements of real-time broadcast monitoring than existing algorithms.Finally,this paper adopts the Linux-equipped Raspberry Pi 4B development board and USB audio capture card,and drives the audio capture card to capture real-time audio streams through the Linux system's Advanced Sound Architecture ALSA and proposes to fuse the cepstrum-based and Shazam-based algorithms for call sign identification,and tests the actual operation effect of the fusion algorithm ported to the embedded system,thus verifying the algorithm's The fusion algorithm was tested on an embedded system,thus verifying the practical value of the algorithm,realising real-time monitoring of broadcast programmes and ensuring the safety of broadcast transmissions.