Screening calls natively in an IP multimedia subsystem network

A non-transitory, computer-readable storage medium comprising instructions recorded thereon, wherein the instructions, when executed by at least one data processor of a system of a Media Resource Function (MRF) configured as a network element disposed in a telecommunications network, cause the system to:

• intercept a first inbound call from a caller to a subscriber of the telecommunications network received at a Telephony Application Server (TAS) communicatively coupled to the MRF;
• redirect the first inbound call to an interactive voice response (IVR) system,
• wherein the IVR is configured to present a voice prompt to the caller to state a name of the caller, and
• wherein the IVR is configured to create an audio recording of a response given by the caller in response to the voice prompt;
• relay the audio recording to the subscriber of the telecommunications network;
• cause a device of the subscriber to present at least a first option to the subscriber to select a disposition for the first inbound call,
• wherein the at least a first option presented to the subscriber includes an option to accept the first inbound call;
• cause the telecommunications network, upon the subscriber selecting the option to accept the first inbound call, to forward the first inbound call to the subscriber;
• and
• cause the system to implement a machine learning (ML) analytics and large language model (LLM), the ML analytics and LLM being caused to:
• upon receiving an eighth inbound call, conduct an audio fingerprinting analysis of an audio sample of a voice of the caller to:
• determine whether the caller has a robotic tone,
• determine whether a waveform of the audio sample matches a waveform of at least one audio sample of a robocall known to an operator of the telecommunications network, or
• determine whether the audio sample matches at least one linguistic pattern of a fraudulent call known to the operator of the telecommunications network;
• and
• wherein the system is caused, upon determining that the caller has a robotic tone, the eighth inbound call is a robocall, or the eighth inbound call is a fraudulent call, reject the eighth inbound call.

A non-transitory, computer-readable storage medium comprising instructions recorded thereon, wherein the instructions, when executed by at least one data processor of a system of a telecommunications network, cause the system to implement a machine learning (ML) analytics and large language model (LLM), the ML analytics and LLM being caused to:

• upon a Media Resource Function (MRF) of the system intercepting a first inbound call from a caller to a subscriber of the telecommunications network received at a Telephony Application Server (TAS) communicatively coupled to the MRF and redirecting the first inbound call to an interactive voice response (IVR) system of the telecommunications network,
• wherein the IVR is configured to prompt the caller to state a purpose of the first inbound call,
• wherein the IVR is further configured to create an audio recording of a response given by the caller to the voice prompt,
• wherein the MRF is configured to relay the audio recording to the subscriber of the telecommunications network,
• wherein the MRF is further configured to cause a device of the subscriber to present at least a first option to the subscriber of the telecommunications network to select a disposition for a fourth inbound call,
• wherein the at least a first option presented to the subscriber includes an option to accept the fourth inbound call, and
• wherein the MRF is further configured to, upon the subscriber selecting the option to accept the fourth inbound call, to forward the fourth inbound call to the subscriber;
• conduct an analysis of at least one word, at least one phrase, or at least one tonality of an audio sample of a response of the caller,
• wherein conducting the analysis of the at least one word, at least one phrase, or at least one tonality of the audio sample further comprises conducting an audio fingerprinting analysis of the audio sample to:
• determine whether the caller has a robotic tone,
• determine whether a waveform of the audio sample matches a waveform of at least one audio sample of a robocall known to an operator of the telecommunications network, or
• determine whether the audio sample matches at least one linguistic pattern of a fraudulent call known to the operator of the telecommunications network;
• determine whether the first inbound call is a spam call or a robocall based on the analysis;
• analyze a purpose of the first inbound call stated by the caller;
• and
• upon determination by the ML analytics and LLM that the first inbound call is a spam call or a robocall, reject the first inbound call.