Communication media exchange method and system based on SIP soft exchange

Abstract

The invention relates to a communication media exchange method and system based on SIP soft switch, and belongs to the technical field of SIP soft switch communication. The method comprises the following steps of: determining a media stream transmission endpoint based on soft switch signaling call routing of SIP (Session Initiation Protocol); processing and distributing the original media stream in real time; and the processed media stream is directly docked to realize intelligent interaction. The method has the effect that the docking efficiency with the AI large model can be improved as much as possible.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of SIP soft switching communication, in particular to a communication media exchange method and system based on SIP soft switching. BACKGROUND

[0002] With the rapid development of communication technology, people's demand for communication is increasingly diversified, not only requiring to realize basic voice call, but also expecting to be able to make high-definition video call, multimedia message transmission, real-time data sharing and other rich communication services. At the same time, with the development of AI large model technology, more and more large model technology is applied to communication, such as intelligent voice assistant, intelligent voice customer service robot, etc. However, the existing SIP-based communication system still has some problems in media exchange. In a complex network environment, with the development of communication services and AI large model technology, the traditional SIP soft switching communication system is complex in interfacing with large model voice, generally based on MRCP protocol interfacing, and the communication mode is complex, and the minimum resource requirement and real-time effect are general. SUMMARY

[0003] In order to solve the problem of low efficiency of interfacing with AI large model as much as possible, the present application provides a communication media exchange method and system based on SIP soft switching.

[0004] On the one hand, the communication media exchange method based on SIP soft switching provided by the present application adopts the following technical solution: A communication media exchange method based on SIP soft switching, comprising the following steps: SIP-based soft switching signaling call routing to determine media stream transmission endpoints; Real-time processing and distribution of original media stream; Directly interfacing with the processed media stream to realize intelligent interaction.

[0005] By adopting the above technical solution, through the optimized signaling call routing mechanism, an efficient media stream transmission channel is established to provide a low-delay media transmission basis for AI interaction; by adopting intelligent media stream processing technology, real-time transcoding and multi-distribution of media stream are realized to meet the efficient processing demand of AI large model for media data; the processed media stream is directly interfaced innovatively, which saves the complex conversion link of traditional MRCP protocol, and significantly improves the real-time performance and fluency of AI voice interaction.

[0006] As a preferred, the SIP-based soft switching signaling call routing to determine media stream transmission endpoints comprises the following steps: Terminal registration, wherein the terminal is set as a system extension, a PSTN gateway or a WebRTC client; Call initiation and number verification; Dynamic routing decisions are made by executing full match rules, regular expression rules, and source information rules in order of priority. Dual-channel media call establishment involves establishing a first media channel with the calling terminal and a second media channel with the called terminal, and then performing intelligent bridging.

[0007] By adopting the above technical solutions, it is ensured that calls can be accurately routed to the target terminal, while the reliability and security of media transmission are improved through dual-channel bridging.

[0008] Preferably, the dynamic routing decision includes: When the exact match rule is hit, the called number is compared digit by digit with the rule number pre-configured in the routing table. If they match exactly, the caller is routed to the specified destination; otherwise, the regular expression rule is applied. When the regular expression rule is matched, the called number is matched with the preset regular expression. If the match is successful, the call is routed to the bound destination. When neither the exact match rule nor the regular expression rule is matched, a match is performed based on the source information in the terminal's sent information.

[0009] By adopting the above technical solution, precise routing is achieved through three-level dynamic routing rules, thereby improving call success rate, reducing invalid routes, and optimizing network resource utilization.

[0010] Preferably, the dual-channel media call setup includes: The softswitch system accepts the call request from the calling terminal as the called party and establishes the first media channel; The softswitch system, based on the routing decision, initiates a call request to the called terminal as the calling party, and establishes a second media channel. Intelligent bridging of the first media channel and the second media channel.

[0011] By adopting the above technical solution, and by establishing a two-way media channel between the caller and the called party and performing intelligent bridging, control over the media stream can be achieved.

[0012] Preferably, the raw media stream is processed and distributed in real time, including the following steps: Cross-terminal media capability negotiation and adaptation: By parsing the media parameters of the calling terminal and dynamically negotiating with the called terminal, the common encoding format supported by the calling terminal and the called terminal is determined. Intelligent media stream bridging and multi-channel distribution establishes a bidirectional media channel bridge, performs real-time transcoding of mismatched encoding formats, and copies and distributes the media stream to the recording server and real-time analysis server.

[0013] By adopting the above technical solutions, the incompatibility of different terminal encoding formats is solved through cross-terminal media capability negotiation and real-time transcoding, ensuring call quality; the multi-channel distribution function supports recording and real-time analysis, expanding the application scenarios of the system.

[0014] As a preferred option, the raw media stream is processed and distributed in real time, and media resource management is also included; media data is temporarily backed up during the call and codec resources and network ports are released after the call ends, and resource utilization is improved through resource optimization algorithms.

[0015] By adopting the above technical solutions, temporary backup and resource release can be achieved through media resource management, thereby optimizing system resource utilization and improving system stability and fault recovery capabilities.

[0016] As a preferred approach, intelligent interaction is achieved by directly connecting to the processed media stream, including the following steps: Intelligent interaction channel initialization and global voice capture: By establishing a two-way connection with the large voice model engine and enabling global sound recording mode, the system continuously captures user voice input. Streaming semantic understanding and real-time speech synthesis convert the captured speech into a text stream in real time and transmit it to the large model engine. At the same time, the generated text response is converted into a speech stream output through the TTS module. For large-scale speech stream processing, a differentiated processing strategy is adopted according to the type of large-scale model. For text-based large-scale models, speech-to-text and then-to-speech conversion is performed, while for end-to-end speech large-scale models, the original speech stream is processed directly.

[0017] By adopting the above technical solutions, intelligent interaction with large speech models is achieved, supporting real-time speech capture, streaming semantic understanding and speech synthesis, thus improving user experience; differentiated processing strategies are adapted to different types of large models to improve interaction efficiency.

[0018] On the other hand, this application also provides a communication media switching system based on SIP softswitch, which adopts the following technical solution: A communication media switching system based on SIP softswitch, comprising: The terminal unit is used to output SIP signaling, call control information, and voice media streams; SIP softswitch unit, used to handle SIP signaling and call routing; The softswitch media processing unit is used to process media streams; Intelligent voice interaction unit, used to realize intelligent voice interaction; The output signal of the terminal unit is connected to the input of the SIP softswitch unit and the softswitch media processing unit, and the output signal of the SIP softswitch unit and the softswitch media processing unit is connected to the input of the intelligent voice interaction unit.

[0019] By adopting the above technical solutions, the modular design enables the coordinated operation of signaling processing, media exchange, and intelligent interaction, supporting efficient and flexible deployment of communication services.

[0020] Preferably, the SIP softswitch unit includes: The SIP signaling processing module is used to process SIP signaling. The call control interface module is used to provide a call control interface; The call routing module is used to execute full match rules, regular expression rules, and source information rules in order of priority to make routing decisions.

[0021] By adopting the above technical solutions, the SIP softswitch unit ensures accurate call routing and efficient processing through signaling processing, call control, and dynamic routing functions, thereby improving the system's reliability and scalability.

[0022] Preferably, the softswitch media processing unit includes: The media stream switching module is used to establish and maintain bidirectional media stream channels; The media encoding conversion module adopts a dual-mode architecture of software DSP and hardware acceleration chip to handle real-time transcoding between different encoding formats; The media stream recording and backup module is used to copy the original media stream during media exchange and push the copied media stream to the recording server and the real-time streaming media analysis server respectively; it also backs up the original media data to provide fault recovery capability, and automatically deletes the backup data to free up storage space after the call is confirmed.

[0023] By adopting the above technical solutions, the softswitch media processing unit ensures high-quality transmission and diversified processing of media streams through media stream switching, real-time transcoding, and multi-path distribution functions; the backup mechanism provides fault recovery capabilities and optimizes storage resource management.

[0024] In summary, this application includes at least one of the following beneficial technical effects: 1. Intelligent interaction is achieved by directly connecting to the processed media stream, eliminating the complex connection method of the traditional MRCP protocol, which greatly improves the real-time performance and fluency of AI voice interaction and enhances the system response speed; 2. Through the collaborative design of dynamic routing rules and dual-channel media bridging mechanism, the connection efficiency of cross-network and cross-terminal calls is significantly improved. At the same time, through real-time media stream transcoding and bridging technology, seamless interoperability of heterogeneous terminals is achieved, greatly improving call quality. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall process of Embodiment 1 of this application; Figure 2 This is a flowchart illustrating step S1 in Embodiment 1 of this application; Figure 3 This is a flowchart illustrating step S2 in Embodiment 1 of this application; Figure 4 This is a flowchart illustrating step S3 in Embodiment 1 of this application; Figure 5 This is a schematic diagram of Embodiment 2 of this application. Detailed Implementation

[0026] The following combination Figures 1-5 This application will be described in further detail.

[0027] Example 1: Embodiment 1 of this application discloses a communication media switching method based on SIP softswitch.

[0028] Reference Figure 1 A communication media switching method based on SIP softswitch includes the following steps: S1: SIP-based softswitch signaling call routing to determine media stream transmission endpoints; S2: Process and distribute the raw media stream in real time; S3: Directly connects to the processed media stream to achieve intelligent interaction.

[0029] Reference Figure 2 Step S1, based on SIP-based softswitch signaling call routing, determines the media stream transmission endpoint, including the following steps: S11: Terminal Registration. The terminal can be configured as a system extension, PSTN gateway, or WebRTC client.

[0030] If the terminal is set as a system extension, during registration, the system extension first sends a REGISTER request to the SIP server, and then the registration service verifies the system extension's account, password, and permissions. After successful verification, the system extension's contact information is written back to the registration database, which may include IP address, port, and protocol.

[0031] If the terminal is set as a PSTN gateway, the number-gateway mapping and registration will be completed by the PSTN gateway.

[0032] If the terminal is set to a WebRTC client, then the NAT address translation and registration will be handled by the WebRTC gateway.

[0033] S12: Call Initiation and Number Verification. The terminal sends an INVITE message to the SIP server, including the called number. After receiving the INVITE message, the SIP server performs message format and authentication verification.

[0034] S13: Dynamic routing decision. The routing service executes three types of matching rules in order of priority: exact match rules, regular expression rules, and source information rules. Once a matching rule is found, the matching of subsequent rules stops.

[0035] Specifically, the exact match rule works by comparing the called number with all pre-configured rule numbers in the routing table when the system needs to route a call. If a rule number exactly matches the called number, the call is immediately routed to the specified destination corresponding to that rule, and no further checks are performed on other rules.

[0036] Regular expression matching: If the called number matches a preset regular expression, the call is directed to the destination bound to that rule.

[0037] Source information matching: If the first two steps fail, a rule comparison is performed again based on the source information (source IP range, User-Agent identifier, terminal type field, etc.) in the INVITE message header. If a match is found, the route is routed according to that rule.

[0038] If the destination terminal is set to a system extension, the call request will be forwarded to the registration server. The registration server will first check its local database to see if the system extension is already registered. If the system extension is registered, the INVITE message will be forwarded to the IP / port recorded during registration. If the system extension is not registered, the system will prompt that the system extension is not registered and not found, and the call will fail.

[0039] If the destination terminal is set to a PSTN gateway, the call request will be forwarded to the PSTN gateway service. The gateway service will then perform secondary routing based on fields such as the called number's operator and location. The best trunk is selected through the local number segment-operator mapping table. If multiple available trunks exist, they can be further selected based on load balancing or rate strategies.

[0040] If the destination terminal is set to a WebRTC client, the call request will be forwarded to the WebRTC gateway service; the gateway will then forward the call based on the locally maintained WebRTC client mapping table.

[0041] S14: Dual-channel media call establishment. When the calling terminal initiates a call, the softswitch system first accepts the request as the called party. At this time, the softswitch system establishes a first media channel with the calling terminal to transmit early media information such as ringback tones. Based on the routing decision, the softswitch system then initiates a new call request to the called terminal as the calling party. At this time, a second media channel is established, and the second media channel is intelligently bridged with the first media channel.

[0042] By establishing a primary and secondary media channel, complete control over the media stream can be achieved through intelligent bridging, providing the infrastructure for transcoding, recording, and AI processing. Furthermore, it effectively hides the called terminal's real network address, enhancing system security. Once both the primary and secondary media channels are successfully established, the system automatically completes bidirectional media stream connection. If the encoding formats at both ends are inconsistent (e.g., the calling party supports G.729 while the called party only supports PCMA), the softswitch will perform transcoding in real time. The entire process is strictly monitored by a state machine; failure at any stage will trigger a corresponding exception handling mechanism (e.g., returning a SIP error code) to ensure timely resource release.

[0043] Reference Figure 3 Step S2 involves real-time processing and distribution of the original media stream, including the following steps: S21: Cross-terminal media capability negotiation and adaptation.

[0044] After a successful call setup, the system parses the media parameters (such as supported audio encoding types) carried in the INVITE message by the calling terminal and selects the optimal common encoding format for confirmation. If it finds that the calling terminal and the system do not support the same encoding format, the call process is immediately terminated to avoid invalid connections. Next, based on the routing decision, the system, acting as the calling terminal, initiates media capability negotiation with the target terminal, dynamically adjusting the proposed parameters according to the called terminal type to ensure a consistent media transmission scheme is reached with the called terminal.

[0045] S22: Intelligent Media Stream Bridging and Multi-path Distribution. The system establishes a bidirectional media channel bridge. When it detects that the calling and called parties are using different encoding formats, the system's built-in media transcoding engine will complete the encoding and decoding conversion of the audio stream in real time, ensuring that the call quality is not affected by encoding differences. At the same time, the system will copy the original media stream and distribute it to multiple auxiliary systems: one path is sent to the recording server for digital storage, and another path is transmitted to the real-time streaming media analysis server to provide a data source for AI voice processing.

[0046] S23: Media Resource Management. During a call, raw media data is temporarily backed up to provide fault recovery capabilities. After the call ends, the system immediately releases the occupied codec resources, network ports, and storage space, and optimizes resource pool utilization through algorithms such as LRU. The entire media exchange process embodies the design principles of high compatibility, scalability, and reliability, laying a solid foundation for upper-layer business functions.

[0047] Reference Figure 4 Step S3 involves directly connecting to the processed media stream to achieve intelligent interaction, including the following steps: S31: Intelligent Interconnection Channel Initialization and Global Voice Capture. When the system detects that a call has been routed to the AI ​​customer service channel, it automatically triggers the large-scale model service initialization process. The system will immediately establish a two-way connection with the voice large-scale model engine and simultaneously activate the global voice capture mode to continuously capture all voice input from the user. During this stage, the system will play a welcome voice prompt according to a preset scenario, such as "Welcome to xxx. How can I help you? You can speak your specific question in Mandarin," laying the groundwork for subsequent interactions.

[0048] S32: Streaming semantic understanding and real-time speech synthesis. After the audio system and speech recognition are integrated, the system receives the user's question and integrates the question text with a large text model to generate the answer text. Simultaneously, a streaming audio media stream is generated, stored, and transmitted to the call channel, at which point the user receives the automatic reply.

[0049] S33: Large Model Speech Stream Processing. The system employs differentiated processing strategies based on the type of large model it connects to. For text-based large models, the system uses a high-precision ASR speech recognition system to convert user speech into a text stream in real time, and then uses streaming technology to send text segments to the large model engine. Simultaneously, the large model can convert the generated text response into a speech stream through a low-latency TTS module, achieving a "speak-while-play" effect. For end-to-end speech large models, the system directly transmits the raw speech stream to the model input and mixes the model's output speech stream into the call channel in real time.

[0050] The implementation principle of a communication media switching method based on SIP softswitch in Embodiment 1 of this application is as follows: This embodiment, by pre-setting multiple routing rules that can be combined, can more accurately route call requests to appropriate terminal service modules in complex network environments, optimizing media transmission paths and reducing latency and packet loss rates. Operations such as encoding / decoding negotiation, format conversion, and traffic replication on the media stream ensure compatibility with different encoding formats, enabling media stream analysis and recording functions, and facilitating the integration of AI functions. A new approach is adopted for interfacing with the large voice model, improving interfacing efficiency and real-time performance, and enhancing user experience. Simultaneously, all these operations are completed within the softswitch system, independent of terminal devices, improving the system's autonomy and reliability, representing a significant improvement and enhancement compared to existing technologies.

[0051] Example 2: This application discloses a communication media switching system based on SIP softswitch.

[0052] Reference Figure 5 A communication media switching system based on SIP softswitch is disclosed, comprising a terminal unit, a SIP softswitch unit, a softswitch media processing unit, and an intelligent voice interaction unit. The output signal of the terminal unit is connected to the input signal of the SIP softswitch unit for outputting SIP signaling and call control information; the output signal of the terminal unit is also connected to the input signal of the softswitch media processing unit for outputting voice media streams. The input signal of the SIP softswitch unit is connected to the input signal of the intelligent voice interaction unit for interactive control; the output signal of the softswitch media processing unit is connected to the input signal of the intelligent voice interaction unit for interactive media interfacing.

[0053] The terminal unit includes a registration service module, a WebRTC gateway service module, and a PSTN gateway service module. The registration service module includes system extensions, the WebRTC gateway service module includes WebRTC extensions, and the PSTN gateway service module includes a PSTN gateway. The registration service module is used for account verification and contact information recording of system extensions. During registration, the system extension first sends a REGISTER request to the SIP server. Then, the registration service verifies the system extension's account, password, and permissions. After successful verification, the system extension's contact information is written back to the registration database. Contact information may include IP address, port, and protocol. The registration service module can be implemented by server-side software; alternatively, a distributed server cluster can be used to complete the registration service.

[0054] The WebRTC gateway service module is used to perform client NAT traversal and registration. When the terminal is set as a WebRTC client, the WebRTC gateway completes NAT address translation and registration. The WebRTC gateway can be a dedicated network device or a software gateway. Replaceable gateway types can also include virtual gateways based on cloud platforms.

[0055] The PSTN gateway service module is used to perform the mapping and registration between numbers and physical gateways. If the terminal is set as a PSTN gateway, the PSTN gateway will complete the number-gateway mapping and registration. The PSTN gateway can be a traditional telecommunications network device, and alternative gateway types can also be new gateways that integrate multiple communication technologies.

[0056] The SIP softswitch unit includes a SIP signaling processing module, a call control interface module, and a call routing module. The signaling processing module is responsible for processing SIP signaling, performing format and authentication checks on signaling messages sent by terminals. This module can be a software program running on a server, or alternatively, a hardware chip can be used to accelerate signaling processing. The routing decision module executes three types of matching rules sequentially according to priority: exact match rules, regular expression rules, and source information rules. Once a matching rule is found, subsequent matching is stopped. The routing decision module can perform hierarchical matching of called numbers based on preset rules, and the three-level routing rule base can store a large number of routing rules.

[0057] The softswitch media processing unit includes a media stream switching module, a media encoding conversion module, and a media stream recording and backup module. The media stream switching module, acting as the basic path controller, is responsible for establishing, maintaining, and dismantling media stream channels, supporting real-time bridging and mixing of bidirectional RTP / RTCP streams, and can flexibly handle one-to-one calls or multi-party conference scenarios. The media encoding conversion module adopts a dual-mode architecture of software DSP and hardware acceleration chip, dynamically detecting encoding differences between terminals (such as G.711 and Opus), and achieving PCM intermediate layer transcoding through a cascaded decoding-re-encoding process, ensuring seamless interoperability between terminals of different standards, with transcoding latency controlled within 5 milliseconds.

[0058] The media stream recording and backup module copies the original media stream during media exchange and pushes the copied media stream to the recording server and the real-time streaming media analysis server respectively, realizing call recording and real-time analysis functions. Furthermore, the media stream recording and backup module backs up the original media data for recovery in case of system failure, and promptly deletes the backup data after call confirmation to free up storage space.

[0059] The intelligent voice interaction unit comprises an ASR (Automatic Speech Recognition) subsystem, an end-to-end large-scale speech model, and a text-based large-scale model. The ASR subsystem converts user speech into text, for example, by using a high-precision ASR module to convert user speech into a text stream in real time. The end-to-end large-scale speech model directly transmits the raw speech stream to its input and mixes the model's output speech stream into the call channel in real time, achieving direct speech stream processing. The text-based large-scale model converts the text response generated by the large-scale model into a speech stream through a low-latency TTS (Text-to-Speech) module, enabling voice interaction with the user.

[0060] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A communication media switching method based on SIP softswitch, characterized in that: Includes the following steps: SIP-based softswitch signaling call routing determines media stream transmission endpoints; Process and distribute raw media streams in real time; It enables intelligent interaction by directly connecting to the processed media stream.

2. The communication media switching method based on SIP softswitch according to claim 1, characterized in that: SIP-based softswitch signaling call routing determines media stream transmission endpoints, including the following steps: Terminal registration, where the terminal is set as a system extension, PSTN gateway, or WebRTC client; Call initiation and number verification; Dynamic routing decisions are made by executing full match rules, regular expression rules, and source information rules in order of priority. Dual-channel media call establishment involves establishing a first media channel with the calling terminal and a second media channel with the called terminal, and then performing intelligent bridging.

3. The communication media switching method based on SIP softswitch according to claim 2, characterized in that: The dynamic routing decision includes: When the exact match rule is hit, the called number is compared digit by digit with the rule number pre-configured in the routing table. If they match exactly, the caller is routed to the specified destination; otherwise, the regular expression rule is applied. When the regular expression rule is matched, the called number is matched with the preset regular expression. If the match is successful, the call is routed to the bound destination. When neither the exact match rule nor the regular expression rule is matched, a match is performed based on the source information in the terminal's sent information.

4. A communication media switching method based on SIP softswitch according to claim 2, characterized in that: The establishment of the dual-channel media call includes: The softswitch system accepts the call request from the calling terminal as the called party and establishes the first media channel; The softswitch system, based on the routing decision, initiates a call request to the called terminal as the calling party, and establishes a second media channel. Intelligent bridging of the first media channel and the second media channel.

5. A communication media switching method based on SIP softswitch according to claim 1, characterized in that: Real-time processing and distribution of raw media streams includes the following steps: Cross-terminal media capability negotiation and adaptation: By parsing the media parameters of the calling terminal and dynamically negotiating with the called terminal, the common encoding format supported by the calling terminal and the called terminal is determined. Intelligent media stream bridging and multi-channel distribution establishes a bidirectional media channel bridge, performs real-time transcoding of mismatched encoding formats, and copies and distributes the media stream to the recording server and real-time analysis server.

6. A communication media switching method based on SIP softswitch according to claim 5, characterized in that: It performs real-time processing and distribution of raw media streams, and also includes media resource management; it performs temporary backups of media data during calls and releases codec resources and network ports after the call ends, and improves resource utilization through resource optimization algorithms.

7. A communication media switching method based on SIP softswitch according to claim 1, characterized in that: To achieve intelligent interaction by directly connecting to the processed media stream, the following steps are included: Intelligent interaction channel initialization and global voice capture: By establishing a two-way connection with the large voice model engine and enabling global sound recording mode, the system continuously captures user voice input. Streaming semantic understanding and real-time speech synthesis convert the captured speech into a text stream in real time and transmit it to the large model engine. At the same time, the generated text response is converted into a speech stream output through the TTS module. For large-scale speech stream processing, a differentiated processing strategy is adopted according to the type of large-scale model. For text-based large-scale models, speech-to-text and then-to-speech conversion is performed, while for end-to-end speech large-scale models, the original speech stream is processed directly.

8. A communication media switching system based on SIP softswitch, applicable to the communication media switching method described in any one of claims 1-7, characterized in that: include; The terminal unit is used to output SIP signaling, call control information, and voice media streams; SIP softswitch unit, used to handle SIP signaling and call routing; The softswitch media processing unit is used to process media streams; Intelligent voice interaction unit, used to realize intelligent voice interaction; The output signal of the terminal unit is connected to the input of the SIP softswitch unit and the softswitch media processing unit, and the output signal of the SIP softswitch unit and the softswitch media processing unit is connected to the input of the intelligent voice interaction unit.

9. A communication media switching system based on SIP softswitch according to claim 8, characterized in that: The SIP softswitch unit includes: The SIP signaling processing module is used to process SIP signaling. The call control interface module is used to provide a call control interface; The call routing module is used to execute full match rules, regular expression rules, and source information rules in order of priority to make routing decisions.

10. A communication media switching system based on SIP softswitch according to claim 8, characterized in that: The softswitch media processing unit includes: The media stream switching module is used to establish and maintain bidirectional media stream channels; The media encoding conversion module adopts a dual-mode architecture of software DSP and hardware acceleration chip to handle real-time transcoding between different encoding formats; The media stream recording and backup module is used to copy the original media stream during media exchange and push the copied media stream to the recording server and the real-time streaming media analysis server respectively; it also backs up the original media data to provide fault recovery capability, and automatically deletes the backup data to free up storage space after the call is confirmed.

Images