Audio processing for multi-participant communication systems

摘要

Audio processing is provided to determine whether an audio issue is present within a multi-participant communication system such as a teleconference or videoconference bridge or a trunk dispatch system. Audio issues such as background noise, background conversations, or other unwanted audio that is being interjected into the multi-participant conversation and that may be dominating the audio are detected by measuring characteristics of audio samples taken from the communication ports of the multi-participant communication system. A correction may then be applied to the audio received through the communication port by a processor of the multi-participant communication system without intervention by an administrator, such as by muting the port, applying a noise cancellation to audio from the port, or time-shifting the audio from the port.

主权项

1. A conference bridge, comprising:
a plurality of conference bridge ports through which audio is delivered to and received from conference participants, the audio from each conference bridge port being shared with the other conference bridge ports of the plurality; a processor that is in communication with the plurality of conference bridge ports; and a non-transitory memory storing instructions which, when executed by the processor, cause the processor to perform operations comprising:
analyzing audio received from each of the plurality of conference bridge ports to determine which conference bridge ports are providing audio that includes characteristics matching at least one criterion by comparing a sample of audio from each conference bridge port, the sample including frequencies, magnitudes, and durations, to a reference pattern of audio, wherein the at least one criterion includes a deviation from the reference pattern of audio;associating a conference bridge participant among a plurality of conference bridge participants with each of the plurality of conference bridge ports;determining a location of each of the conference bridge participants; andapplying a correction to the audio from each of the conference bridge ports that includes the characteristics matching the at least one criterion prior to the audio being shared with the other conference bridge ports of the plurality, wherein the correction being applied by the processor varies based on the conference bridge port producing the audio being corrected, based on the conference bridge participant associated with the conference bridge port producing the audio being corrected, and based on the location of the conference bridge participant associated with the conference bridge port producing the audio being corrected;wherein the trunk dispatch processor applies the correction by muting and recording a period of audio from a first conference bridge port when the audio from the first conference bridge port overlaps with audio from a second conference bridge port, and replaying the recording of the audio to the plurality of conference bridge ports upon detecting an end of a period of audio from the second conference bridge port.