Research And Implementation Of An Algorithm For Tone Detection Based On Voice Service Of VOIP

Voice service of VOIP, which delivers voice signals and completes voicetelephone calls through IP networks, has become one of the popular commercialcommunication technologies. VOIP gateway achieves the interoperability betweenthe public switched telephone network and the IP networks, thus makes it possiblefor the traditional telephones to access the IP networks to make calls as well. Tonedetection is an important module of VOIP gateway, which is applied to manyprocesses, such as services switching and number identification. At present, due tothe large numbers of types and quantities for tones which are required to detected,the main approach to detect different tones is the use of a variety of specializedchips. This kind of circuit design is not only lack of flexible and high cost, but alsohas a low detecting accuracy and poor anti-noise performance. By the research ondifferent tones to be detected, a tone detection algorithm using software is designedand implemented in this paper, which makes the unified detection of different tonescome true.The overall architecture of the algorithm using software is determinedaccording to the characteristics of the tones to be detected in VOIP voice service.The detection threshold of signal tones is given by defining the parameter table, andthe algorithm is consisted of two modules, that is, the fast Fourier transform andpre-processing module and the sound judgment module. The parameter table can beadded or deleted freely in order to give a flexible support to the detection ofdifferent tones. The frequency domain information of signals is calculated andextracted by Fast Fourier transform and the preprocessing module, after that, thetype of the signal can be identified. Sound judgment module determines the durationof the signal mainly by its energy in time domain, and then reports the signal typeafter identifying it. The performance of the algorithm is optimized in order to meetthe requirements of network testing and commercial use. Through the improvementof the search method of signal energy in time domain, and the set of flexible SNRthreshold, the signal detection index in protocol is achieved, and at the same time,the anti-voice and noise immunity performances of the dual-tone multi-frequencysignal are compatible. Complete test program for the algorithm is compiled andperformed, detailed performance test report and analysis of the test results are givenin order to verify the performance of the algorithm.The overall design, optimization and test of tone detection algorithm iscompleted in this paper, it passes a rigorous and comprehensive test program. Theunified detection for dozens of agreement sounds (mono or combination of sound) and dual-tone multi-frequency signals is realized by this algorithm, at the samemoment, the algorithm is able to add or delete the supported types of tonesaccording to service requirement. Test results verify that this algorithm fully meetthe test index of dual-tone multi-frequency signals, and the performances of anti-noise, anti-voice and anti-low-frequency-high-energy are excellent, furthermore, thecompatibility of different performances is realized. This algorithm has beensuccessfully used in the next-generation featured gateway of a company.