Enhanced Multiple Speakers’ Separation and Identification for VOIP Applications Using Deep Learning

This project is to simulate the Voice over Internet Protocol (VoIP) over IEEE 802.11b and 802.11e WLANs (Wireless Local Area Network) with different Coder-Decoder (CODECs) by using Network Simulator-2 (NS-2). The main objectives of this project are to analyze the VoIP of Mean Opinion Score (MOS), throughput and packet drop rate. The simulation is conducted using different CODECs (G.711, G.729A and G.723.1) and simulated without background traffic (data). The simulation in this project is using 1 wired node and 1 access point connected with mobile nodes in hierarchy. The mobile nodes will send VoIP calls to wired node through access point simultaneously. From the project, it is found that the CODEC G.711 provides the best quality of VoIP calls. However, CODEC G.723.1 can support the most VoIP calls with acceptable quality. In addition, with the first priority given by 802.11e to the VoIP, it can provide a very good perceived audio quality.

This paper explores the opportunities of using not standardized compression codecs while transmitting the voice data over the existing network infrastructure of VoIP (Voice Over Internet Protocol) providers based on SIP (Session Initiation Protocol) in the Internet. Telecommunication networks offer a large set of services aimed at satisfying the ever increasing needs of users. One of them is a VoIP service implemented through network IP/TCP or IP/UDP stack. One of the most important determinants of this process is the need to use data compression. Codec implementation allows you to manage the data compression degree by changing the width of bandwidth usage and voice quality modeling. These features have priority when using VoIP on mobile devices on pre-paid cards such as smartphones where constraints are the size of the packet, the area signal strength in your area and the computing power of the device. Therefore, it is reasonable to develop new technology that provides flexibility in this regard. First of all, implementation of new codecs and ability to put them into VoIP calls according to SIP signaling standards influence on lowering the cost of deployment and then final costs of calls from subscribers located anywhere around the world with access to the Internet even at low bandwidth. The second advantage is the lower network load affecting the reduction of the costs of its maintenance by the owners of the network or service providers.

The increasing use of Voice over Internet Protocol (VoIP) technology in telecom fraud has become a serious global concern. Its ability to spoof caller IDs and IP addresses, and the use of overseas or anonymized servers make VoIP-based scams difficult to trace and regulate. As a result, distinguishing VoIP calls from conventional mobile phone calls based on voice signal characteristics is crucial for enhancing anti-fraud measures. However, existing forensic techniques often struggle to accurately identify speech transmitted via VoIP. To address this challenge, we propose a dual-level 1D-CNN that leverages both frame and utterance features for effective VoIP detection. After evaluating a range of acoustic features, we primarily focus on short-frame Mel-Frequency Cepstral Coefficients (MFCCs) due to their effectiveness in capturing VoIP characteristics. Given the frame-based processing and transmission nature of VoIP, we employ a 1D-CNN, rather than the more commonly used 2D-CNN that treats spectrograms as image, to extract frame-level codec features. Finally, we propose a dual-level classification strategy: the frame-level classifier captures encoding discrepancies within individual frames, while the utterance-level classifier aggregates these frame-level features to learn global encoding patterns through global covariance pooling. Experimental results on the VoIP Phone Call Identification Database (VPCID) demonstrate that the proposed method consistently outperforms existing approaches, delivering superior accuracy and robustness across a wide range of challenging scenarios. Moreover, comprehensive ablation studies validate the effectiveness and rationale behind the design of the proposed model architecture.

Voice over Internet Protocol (VoIP) is no longer only feasible in wired networks, but also applicable through wireless networks as prevalence of wireless networks grows with the technology maturity and rising demand. While making a VoIP call over WiFi wireless networks is applicable, this is not a service with guaranteed quality. With limited bandwidth, an overloaded AP may cause wireless users to experience delayed packets, high packet loss rate, and even low throughput. IEEE 802.11 wireless Local Area Network (WLAN) was originally designed only for data traffic transmission, and therefore deficiency in real-time traffic transmission, such as VoIP applications, is inevitable. Not being able to timely transmit real-time packets not only fails to meet the demand of users, but also increases unnecessary loading in the wireless network. Industries and academies have made lots of efforts on providing QoS for real-time voice traffic which usually coexists with other classes of traffic. In this thesis, a call admission control mechanism is proposed which not only timely monitors the wireless networks, but also dynamically adjusts the number of admissible clients to wireless networks. New VoIP users not only know the network condition before entering it, but also accurately predict transmission quality through sending probing packets. With the mechanism, existing VoIP users in the wireless networks have their service protected, and potential VoIP users can expect good quality of service once admitted.

Information technology is developing rapidly, especially in the field of computer science, telecommunications and information systems. In the early days of telecommunications, voice networks could not be combined with data networks. However, more recently it became possible to integrate data, audio and video services onto a single network, known as a multimedia network. The capabilities of the Internet are also rapidly increasing, and now, in addition to data packets (email, web browsing) the Internet transfers audio and video packets. Voice over Internet Protocol (VoIP) is a multimedia service. This technology can be used to transmit audio, video, and data packets. Consequently, it is becoming the most preferred communications technology. VoIP will eventually replace the use of traditional telephony – Public Switched Telephone Networks (PSTN), although the switchover process is challenging. A chief concern it that, while PSTN transmits audio packets over a closed network, VoIP sends audio packets using an open network – the Internet. In PSTN, eavesdroppers must have direct access to physical network to obtain information from communications. Whereas with VoIP, eavesdroppers can monitor data packets from they are connected to the Internet. Hence, security and privacy are important considerations when switching to VoIP systems. One solution is anonymous systems, which can be used to implement security and privacy protocols. However, this typically reduces the Quality of Service (QoS) of the VoIP. This empirical research study examines VoIP performance in an anonymous network – The Onion Routing (Tor). Two scenarios are implemented using the real Tor network to investigate three QoS metrics for VoIP: latency, jitter and packet loss. As recommended, latency in VoIP should be less than 400 ms, jitter should be less than 50 ms and packet loss should not be more than 5%. In addition, the research calculates the probability of attackers in the two scenarios implemented and evaluate Tor network forecasting. Experiments were conducted in reference to two scenarios. The first scenario is VoIP calls routed through a Tor network with three Tor relays (default Tor). And the second scenario is VoIP calls routed through a Tor network with two Tor relays. Experiments were performed in three periods; December 2012, July 2013, and October 2013. During each experimental time period, a hundred calls were captured for each scenario. Experimental results show that the QoS of VoIP over the two Tor relays was better than the VoIP over the three Tor relays. However, a probability of attackers calculation found that the VoIP with three Tor relays returned better anonymity than that with two Tor relays. The best results were returned over the experimental period in July 2013, when the acceptable calls using two Tor relays reached 64 calls at 5% packet loss. Meanwhile, the worst experimental results were returned in October 2013, when acceptable calls numbered 11 using three Tor relays at 1% packet loss. At the end of 2013, the actual data for relay numbers and bandwidth approached forecast result with time series analysis. Meanwhile, the numbers of Tor users differed from the Tor users forecast. In August 2013, Tor users increased dramatically; this is attributed to the fact that BotNet attack was using the Tor network to attack their target leading to a fivefold increase.

The usage of voice over IP (VoIP) traffic in IEEE 802.11 wireless networks is expected to increase in the near future due to widely deployed 802.11 wireless networks and VoIP services on fixed lines. However, the quality of service (QoS) of VoIP traffic in wireless networks is still unsatisfactory. In this thesis, I identify several sources for the QoS problems of VoIP traffic in IEEE 802.11 wireless networks and propose solutions for these problems. The QoS problems discussed can be divided into three categories, namely, user mobility, VoIP capacity, and call admission control. User mobility causes network disruptions during handoffs. In order to reduce the handoff time between Access Points (APs), I propose a new handoff algorithm, Selective Scanning and Caching, which finds available APs by scanning a minimum number of channels and furthermore allows clients to perform handoffs without scanning, by caching AP information. I also describe a new architecture for the client and server side for seamless IP layer handoffs, which are caused when mobile clients change the subnet due to layer 2 handoffs. I also present two methods to improve VoIP capacity for 802.11 networks, Adaptive Priority Control (APC) and Dynamic Point Coordination Function (DPCF). APC is a new packet scheduling algorithm at the AP and improves the capacity by balancing the uplink and downlink delay of VoIP traffic, and DPCF uses a polling based protocol and minimizes the bandwidth wasted from unnecessary polling, using a dynamic polling list. Additionally, I estimated the capacity for VoIP traffic in IEEE 802.11 wireless networks via theoretical analysis, simulations, and experiments in a wireless test-bed and show how to avoid mistakes in the measurements and comparisons. Finally, to protect the QoS for existing VoIP calls while maximizing the channel utilization, I propose a novel admission control algorithm called QP-CAT (Queue size Prediction using Computation of Additional Transmission), which accurately predicts the impact of new voice calls by virtually transmitting virtual new VoIP traffic.

SummaryNowadays social media plays a key role in interpersonal communication. Instant messaging applications, such as WhatsApp, are used by billions of users every day. Such application has recently been upgraded to support voice calls, whose traffic is transported over the top of network operators, as many other Voice over Internet Protocol (VoIP) services. In this work, we have characterized WhatsApp voice calls through blind traffic detection, which allows to differentiate WhatsApp calls from other applications, such as sharing video, photo, or messaging. Traditional techniques for detection of VoIP calls cannot be applied to the WhatsApp case because the traffic is obfuscated. Actually, our proposal only takes into account the WhatsApp traffic stream statistical features, and not the packet content. From the operators' point of view, the benefits of such a tool are manifold: from traffic prioritization to bespoke marketing campaigns to users that heavily produce WhatsApp voice calls.

Residual networks for text-independent speaker identification: Unleashing the power of residual learning

Mobile and wireless technologies provide Voice over Internet Protocol (VoIP) service to the users with the help of internet. However, there are some limitations with respect to bandwidth, delay, jitter, etc. While it meets congestion and heavy load on the network, it drastically consumes and wastes the resources unnecessarily. Likewise, Mobile Ad-hoc Networks (MANETs) is a part of ad hoc network to provide VoIP communication among the users without the aid of any internet facilities. Since the nodes are mobile in MANETs, it faces many challenges like dynamic topology, limited bandwidth and battery power in order to provide Quality of Service (QoS). This paper proposes a CBT-VoIP technique for reducing the VoIP transmission delay. Ultimately, it helps to improve QoS in MANETs.

Voice over IP (VoIP) is a developing technology and a key factor in both the emerging cyberspace engineering and also an accomplishment to set up its position in the telecom industry. VoIP technology is based on internet technology; where data packets switching system is used rather than circuit switching. Whereas, an analog signal is changed over into the digital signals in the full-duplex transmission. VoIP technology is replaced with the conventional public switched telephone network (PSTN) system due to the high flexibility and low-cost. The purpose of this research work is to deliberate experimental and computational performance in the view of quality of service (QoS) parameters of VoIP over local area network (LAN) network. The VoIP systems implementation is based on two different operating system framework (Linux and Windows), whereas, Linux-based and Windows-based private branch exchanges (PBXs), such as Asterisk (Linux-open source) and Axon (window-close source) are configured, installed and verified. QoS factors (such as packet loss, delay, jitter, etc.) are observed over the Asterisk and Axon PBXs in a LAN domain with the assistance of Paessler switch activity grapher (PRTG) monitoring tool. The validations of results are looked at for QoS parameters crosswise over both PBXs with data load (i.e., file transfer and HTTP traffic) during VoIP calls. The productivity and execution of Axon and Asterisk have been equated and analyzed over experimental based outcomes.

The lecture will present a test of the VoIP (Voice over Internet Protocol) network at the University of Applied Sciences (FHTW) Berlin based on Cisco VoIP phones model 7970G as an example in test of security policy of VoIP networks to find out the security gaps and to support improving security of the FHTW's VoIP campus network. Configuration options of the VoIP phone used at the University of Applied Sciences Berlin will be presented. The finding, installing and configuring products for testing the security of VoIP network and VoIP phones have to be dicussed and the security test of Cisco VoIP phones presented offer ways for make more secure configuration and steps to improve the overall security of VoIP networks. The presented results are backed by the long term ERASMUS TN co-operation between the University of Applied Sciences Berlin and the University of Rousse.

Systems employing the Internet and Voice over Internet Protocol (VoIP) are alternative to the legacy landline telephony system. The services offered by these systems allow users to communicate with their family members, friends, banks and business partners whenever they are online. Particularly, the VoIP service is popular with the Internet users because charges are bound into usually fixed access costs making the price of the long distance calls themselves appear economical and even free. However, VoIP services have several disadvantages. One of the disadvantages of VoIP is, it is exposed to Internet security vulnerabilities, threats and attacks. The attacks come in many forms. One of the methods to attack the VoIP service, is by sending a Spam over Internet Telephony (SPIT). SPIT is normally executed by malicious parties initiating automated, unsolicited, and unwanted communications that used VoIP or video conferencing services, just like an email SPAM. Solutions to mitigate SPIT are still lacking. To defense a VoIP system from SPIT, we have come out with the idea of Automated Intrusion Detection and Prevention System over SPIT (AIDPoS) that can detect and prevent SPIT from external parties to enter the VoIP network. AIDPoS is a combination system of Intrusion Detection System (IDS) and Intrusion Prevention System (IPS) on Voice over IP (VoIP) server. This is an experimental research to find a solution to combat SPIT if it happens in the future. Our result shows that an automated Intrusion detection and Prevention system over SPIT attack is feasible. Our work is on an open source VoIP network. Other vulnerabilities and threats are important, but beyond our research scope and is not covered in this paper.

Voice over Internet protocol (VoIP) is the transmission of voice over networks using the internet protocol. IP networks have become increasingly popular in the past few years, due to the exponential growth of the public internet leading the way in to the IP world. Long distance calls, especially international subscriber dialing (ISD), can be made significantly less expensive when supported by an IP network rather than by the PSTN. Any call made is supported by VoIP technology and it involves the transmission of many individual packets over an IP network. Thus the cost of VoIP calls in part depends on the number and size of packets that must be transmitted. So voice (source) compression technology is used to reduce the amount of bandwidth required in order to reduce cost and to reduce the delay impact from network. But, compression techniques increase the network impairments also. When packets are transmitted through network they are affected by impairments, like packet drop and end-to-end delay. The main agenda of VoIP service providers is to provide good quality of service (QoS). The objective of this research work done is to minimize, the error introduced in channel due to the above mentioned network impairments and hence improve the quality of sound, and also analyze the voice quality with and without use of channel coding scheme. There are two measuring techniques, subjective and objective. In subjective method we measure the mean opinion score (MOS) and use matching algorithms in objective methods to measure the difference between decoded outputs of source coding system and joint source-channel coding system. In this work, the objective method is used and it is observed that the performance of the system improves significantly

In order to provide the same or better service quality in the Internet than traditional circuit-switched telephone network, a number of issues have to be solved which have hampered it in the Internet. Therefore, in this paper the VoIP (Voice over Internet Protocol) network design problems are modeled as nonlinear non convex combinatorial integer mathematical programming problems. The optimization problems have great practice value for the network service providers. This problem is a constrained multicast QoS routing and is proved to be a NP complete problem. The total link capacity augmented cost should be minimized when we design a new VoIP network or when the original network could not serve all of the traffic demands. It is also known as a complicated problem. Introduction Originally, the Internet was designed to provide best-effort services for the data generated by computers. The timeliness of data communication is generally delay tolerant. Quality-of-Service (QoS) constraints are not as important as routing flexibility and connectivity. Hence, using IP to transport voice data is contradictory to the basic requirement of the voice service: a timely delivery of voice samples. Although IP was not initially designed to provide services for real-time traffic, recent technical progress has made IP have the capabilities to provide real-time services in near future. In order to provide the same or better service quality in the Internet than traditional circuit-switched telephone network, we must deal with a number of issues that have hampered it in the Internet. Voice service requirements could be discussed from two perspectives: (1) application requirements such as end-to-end delay, jitter, packet loss and overdue probability; (2) user’s perspective such as reliability, availability, and supplementary services [1][2]. Telephony service providers must guarantee the quality of service they provide e.g. maximum one way delay does not exceed 150ms. Internet telephony requires a range of protocols, ranging from those needed for transporting real-time data across to the network e.g. Real-time Transport protocol (RTP) [3], to Quality-of-Service aware routing (QoS routing), signaling protocol, resource reservation, internetworking between IP networks and PSTN, QoS-aware network management and billing protocols. ITU-T defined H.323 to provide multimedia communication in packet networks. From the perspective of network service providers, they want to optimize the network performance such as minimizing the total bandwidth consumption, maximizing throughput or total revenue [4][5] subject to user and application constraints. In this paper, we want to develop the mathematical model for VoIP network. We minimize the total bandwidth consumption under users’ QoS requirements, the network topology and the network capacity. Performance Optimization Model It is generally accepted that Internet telephony and traditional circuit-switched telephony will coexist for quite some time. The VoIP architecture must deal with interworking between IP networks and PSTN, so we need gateways between the two worlds. There are four possible models of VoIP [6]. They are PC-to-PC, Gateway-to-Gateway, PC-to-Gateway, and Gateway-to-PC models. The architecture of VoIP is shown in Fig.1. The first model of VoIP is PC-to-PC architecture, which based 2nd International Conference on Electrical, Computer Engineering and Electronics (ICECEE 2015) © 2015. The authors Published by Atlantis Press 1144 on the assumption that two or more users have access to multimedia computers that are connected to the Internet. Fig.1 VoIP Architecture The VoIP system is modeled as a graph, where the hosts, VoIP gateways, and switches are represented by nodes and communication link sets are represented by links. Let N = {1, 2,..., n} be the set of nodes and L be the set of links in the graph (network). Let G be the set of all user groups. An user group g is a voice communication session requesting for transmission in the network. An user group g may be an unicast from the source to a destination or a multicast from the source to multiple destinations. For each user group g, the traffic is transmitted exactly over one tree. Dg represents the set of destinations of user group g. g λ is the required bandwidth of the voice transmission for each user group g. Below is a verbal description of the VoIP system design problem we considered. Given : Network topology Capacities of network links Equivalent bandwidth of each multicast/unicast group Time threshold and tolerable overdue probability for each multicast/unicast group To determine : The minimum overall bandwidth consumption Routing tree for each multicast/unicast group Overdue probability of each multicast/unicast group Objective : To minimize the total bandwidth consumption Subject to : End-to-end QoS (overdue probability) constraint Tree constraint Multi-commodity flow constraint Capacity constraint Integrality constraint Hop constraint Table 1 and 2 are the notations we use in this paper . Internet VoIP Gateway VoIPGatewa Central office switch or local hub Central office switch or local hub Multimedia PC IP Router Multimedia PC IP Router Gatekeepe r

LTE (Long Term Evolution), a 4G mobile communication standard has a support for higher data rates and lower latency. Since LTE is completely an Internet Protocol (IP) based technology, Network operators of LTE can handle voice via a “fall-back” technique by using previously existed circuit switched networks. To avoid this, Voice over IP (VoIP) services need to be used to deliver the voice over IP networks, such as Internet. Voice Codec is one of the critical components in VoIP used to convert analog signals to compressed digital packets and vice-versa that can be sent over the internet. But there are different kinds of codecs exist that optimize the voice stream based on bandwidth and application requirement. Unfortunately, there is no standard codec that is acceptable for all circumstances, some codecs provides better quality, some other uses lesser bandwidth. All these features will effect on the ultimate selection of codec by an organization. This paper focuses on the quality of service (QOS) analysis of VoIP traffic over LTE by configuring different voice codecs such as G.711, G.722.2, G.723.1 and G.729. To analyze the performance of these codecs ns-3 simulator is being used since its core has a support for LTE models. The comparative study of above selected VoIP codecs over LTE network in terms of Packet Loss, End to End delay, Jitter, and Throughput in normal and error prone scenarios has been done. From the analyzed results and the performance evaluated, the selection of codec will become easy to provide better VoIP services for customer satisfaction.