Profanity filter for collaboration sessions in heterogenous computing platforms

A memory coupled to a heterogeneous computing platform, wherein the heterogeneous computing platform comprises a plurality of devices configured to communicate via a communication bus through at least one firmware service to firmware service Application Programming Interface (API) independent of any host processor, wherein the memory is configured to receive a plurality of sets of firmware instructions, wherein each set of firmware instructions, upon execution by a respective microcontroller of at least one device of the plurality of devices in a respective microcontroller memory space independent of a host processor memory space wherein the host processor is configured to execute at least an Operating System (OS), configures the plurality of devices to provide a respective plurality of firmware services, wherein at least one device of the plurality of devices operates is configured as an orchestrator configured to perform operations that comprise:

• receive, from another device of the plurality of devices via the at least one firmware service to firmware service API without any involvement by the OS or any host processor, context information that comprises a user presence indication;
• modify a profanity filter setting usable during a collaboration session to enforce a received policy, based, at least in part, upon the context information;
• receive telemetry from at least one device of the plurality of devices via the at least one firmware service to firmware service API without any involvement by the OS or any host processor;
• determine, based at least in part on the telemetry, whether a collaboration session is detected, based at least in part on detection of a pattern of utilization of the at least one device that would indicate a collaboration session is active;
• and
• in response to a determination the collaboration session is active:
• receive a combined audio stream that comprises audio from a plurality of remote participants in the collaboration session;
• send a message, via the at least one firmware service to firmware service API without any involvement by the OS or any host processor, to cause another firmware service to execute at least one algorithm by at least an audio Digital Signal Processor (aDSP) configured to:
• separate speech or utterances produced by the plurality of remote participants in the combined audio stream into a respective plurality of separated individual audio streams, based at least in part on a speaker source separation algorithm applied to the combined audio stream to generate the separated individual audio streams;
• determine whether profanity is detected in each separated individual audio stream of the plurality of separated individual audio streams, based at least in part on a match between at least one of the separated individual audio streams and a set of profanity audio signatures;
• and
• in response to a determination profanity is detected in at least one separated individual audio stream, send a message via the at least one firmware service to firmware service API without any involvement by the OS or any host processor, to cause another firmware service to replace the detected profanity in the at least one separated individual audio stream with another word or utterance to create at least one profanity filtered separated individual audio stream;
• and
• provide the collaboration session with access to a combined profanity filtered audio stream that comprises a plurality of profanity filtered separated individual audio streams.

A method comprising:

• receiving, by an Information Handling System (IHS), a firmware installation package comprising extension drivers in a heterogeneous computing platform of the IHS, wherein the heterogeneous computing platform comprises a plurality of devices configured to communicate via a communication bus through at least one firmware service to firmware service Application Programming Interface (API) independent of any host processor, wherein upon execution by a respective microcontroller of at least one device of the plurality of devices in a respective microcontroller memory space independent of a host processor memory space wherein the host processor is configured to execute at least a host Operating System (OS), each extension driver causes a respective device to provide a firmware service to other devices of the plurality of devices via the at least one firmware service to firmware service API;
• delivering an indication of a policy to an orchestrator among the plurality of devices, wherein the policy is usable by the orchestrator to filter out profanity during a collaboration session using the plurality of devices via the at least one firmware service to firmware service API, without any involvement by any host processor or host OS;
• selecting a profanity filter setting usable during the collaboration session based, at least in part, upon the policy;
• receiving a combined audio stream that comprises audio from a plurality of remote participants in the collaboration session;
• sending a message, from the orchestrator via the at least one firmware service to firmware service API without any involvement by the host OS or any host processor, to cause another device of the plurality of devices to separate speech or utterances produced by the plurality of remote participants in the combined audio stream into a respective plurality of separated individual audio streams, using at least one digital signal processing algorithm comprising a speaker source separation algorithm applied to the combined audio stream to generate the separated individual audio streams;
• determining whether profanity is detected in each separated individual audio stream of the plurality of separated individual audio streams, using the at least one digital signal processing algorithm and based at least in part on determining whether there is a match between at least one of the separated individual audio streams and a set of profanity audio signatures;
• in response to determining profanity is detected in at least one separated individual audio stream, sending a message via the at least one firmware service to firmware service API without any involvement by the host OS or any host processor, to cause another device of the plurality of devices to replace the detected profanity in the at least one separated individual audio stream with another word or utterance to create at least one profanity filtered separated individual audio stream, using the at least one digital signal processing algorithm;
• and
• providing the collaboration session with access to a combined profanity filtered audio stream that comprises a plurality of profanity filtered separated individual audio streams.