Artificial latency for moderating voice communication
A computer-implemented method to determine whether to introduce latency into an audio stream from a particular speaker includes an audio stream from a sender device. The method further includes providing, as input to a trained machine-learning model, the audio stream and a speech analysis score, information about one or more voice emotion parameters, and one or more voice emotion scores for a first user associated with the sender device, wherein the trained machine-learning model is iteratively applied to the audio stream and wherein each iteration corresponds to a respective portion of the audio stream. The method further includes generating as output, with the trained machine-learning model, a level of toxicity in the audio stream. The method further includes transmitting the audio stream to a recipient device, wherein the transmitting is performed to introduce a time delay in the audio stream based on the level of toxicity.
1. A computer-implemented method comprising:
receiving, from a sender device, an audio stream that is directed to a receiver device, wherein the sender device and the receiver device participate in a virtual metaverse;
providing, as input to a trained machine-learning model, the audio stream and a speech analysis score for a first user associated with the sender device;
generating as output, by the trained machine-learning model, a level of toxicity in the audio stream;
identifying silence or a pause between words in the audio stream, the silence or the pause corresponding to a particular timestamp in the audio stream; and
transmitting the audio stream to the receiver device, wherein the transmitting is performed to introduce a time delay in the audio stream based on the level of toxicity and wherein the time delay is introduced as a gap in the audio stream at the particular timestamp of the silence or the pause between words.
2. The method of claim 1 , further comprising:
performing retroactive speech analysis of the audio stream; and
after transmitting the audio stream to the receiver device, providing results of the retroactive speech analysis as input to the trained machine-learning model, wherein generating the level of toxicity by the trained machine-learning model is further based on the results of the retroactive speech analysis.
3. The method of claim 1 , further comprising:
providing a listener history and metadata that describes a reaction of a listener associated with the receiver device to prior instances of toxicity to the trained machine-learning model, wherein generating the level of toxicity by the trained machine-learning model is further based on the listener history and metadata.
4. The method of claim 1 , further comprising:
providing information about one or more voice emotion parameters to the trained machine learning model, the one or more voice emotion parameters is selected from a group of an emotional state of the first user based on an emotion quadrant, a pitch of the audio stream, a vocal effort level as a function of time, and combinations thereof;
wherein the level of toxicity is further based on the one or more voice emotion parameters.
5. The method of claim 1 , wherein the audio stream is part of a video stream, and the method further comprises:
analyzing the audio stream to identify an instance of toxicity;
detecting a portion of video frames in the video stream that depict an avatar associated with the first user performing an action that corresponds to the instance of toxicity in the audio stream; and
responsive to detecting the portion of video frames in the video stream that depict the action, modifying at least the portion of the video frames in the video stream by one or more of blurring the portion or replacing the portion with pixels that match a background region.
6. The method of claim 5 , wherein the action is mouth movement of the avatar that forms words to correspond to the instance of toxicity in the audio stream.
7. The method of claim 1 , wherein the audio stream is provided with a visual signal, and the method further comprises synchronizing the visual signal to the audio stream by introduction of a same time delay in the visual signal as the time delay of the audio stream.
8. A sender device comprising:
a processor; and
a memory coupled to the processor, with instructions stored thereon that, when executed by the processor, cause the processor to perform operations comprising:
receiving an audio stream that is directed to a receiver device, wherein the sender device and the receiver device participate in a virtual metaverse;
providing, as input to a trained machine-learning model, the audio stream and a speech analysis score for a first user associated with the sender device;
generating as output, by the trained machine-learning model, a level of toxicity in the audio stream;
identifying silence or a pause between words in the audio stream, the silence or the pause corresponding to a particular timestamp in the audio stream; and transmitting the audio stream to the receiver device, wherein the transmitting is performed to introduce a time delay in the audio stream based on the level of toxicity and wherein the time delay is introduced as a gap in the audio stream at the particular timestamp of the silence or the pause between words.
9. The sender device of claim 8 , wherein the operations further include:
performing retroactive speech analysis of the audio stream; and
after transmitting the audio stream to the receiver device, providing results of the retroactive speech analysis as input to the trained machine-learning model, wherein generating the level of toxicity by the trained machine-learning model is further based on the results of the retroactive speech analysis.
10. The sender device of claim 8 , wherein the operations further include:
providing a listener history and metadata that describes a reaction of a listener associated with the receiver device to prior instances of toxicity to the trained machine-learning model, wherein generating the level of toxicity by the trained machine-learning model is further based on the listener history and metadata.
11. The sender device of claim 8 , wherein the operations further include:
providing information about one or more voice emotion parameters to the trained machine learning model, the one or more voice emotion parameters is selected from a group of an emotional state of the first user based on an emotion quadrant, a pitch of the audio stream, a vocal effort level as a function of time, and combinations thereof;
wherein the level of toxicity is further based on the one or more voice emotion parameters.
12. The sender device of claim 8 , wherein the audio stream is part of a video stream, and the operations further comprise:
analyzing the audio stream to identify an instance of toxicity;
detecting a portion of video frames in the video stream that depict an avatar associated with the first user performing an action that corresponds to the instance of toxicity in the audio stream; and
responsive to detecting the portion of video frames in the video stream that depict the action, modifying at least the portion of the video frames in the video stream by one or more of blurring the portion or replacing the portion with pixels that match a background region.
13. The sender device of claim 12 , wherein the action is mouth movement of the avatar that forms words to correspond to the instance of toxicity in the audio stream.
14. The sender device of claim 8 , wherein the audio stream is provided with a visual signal, and the operations further comprise synchronizing the visual signal to the audio stream by introduction of a same time delay in the visual signal as the time delay of the audio stream.
15. A non-transitory computer-readable medium with instructions stored thereon that, when executed by one or more computers, cause the one or more computers to perform operations, the operations comprising:
receiving, from a sender device, an audio stream that is directed to a receiver device, wherein the sender device and the receiver device participate in a virtual metaverse;
providing, as input to a trained machine-learning model, the audio stream and a speech analysis score for a first user associated with the sender device;
generating as output, by the trained machine-learning model, a level of toxicity in the audio stream;
identifying silence or a pause between words in the audio stream, the silence or the pause corresponding to a particular timestamp in the audio stream; and
transmitting the audio stream to the receiver device, wherein the transmitting is performed to introduce a time delay in the audio stream based on the level of toxicity and wherein the time delay is introduced as a gap in the audio stream at the particular timestamp of the silence or the pause between words.
16. The non-transitory computer-readable medium of claim 15 , wherein the operations further include:
performing retroactive speech analysis of the audio stream; and
after transmitting the audio stream to the receiver device, providing results of the retroactive speech analysis as input to the trained machine-learning model, wherein generating the level of toxicity by the trained machine-learning model is further based on the results of the retroactive speech analysis.
17. The non-transitory computer-readable medium of claim 15 , wherein the operations further include:
providing a listener history and metadata that describes a reaction of a listener associated with the receiver device to prior instances of toxicity to the trained machine-learning model, wherein generating the level of toxicity by the trained machine-learning model is further based on the listener history and metadata.
18. The non-transitory computer-readable medium of claim 15 , wherein the operations further include:
providing information about one or more voice emotion parameters to the trained machine learning model, the one or more voice emotion parameters is selected from a group of an emotional state of the first user based on an emotion quadrant, a pitch of the audio stream, a vocal effort level as a function of time, and combinations thereof;
wherein the level of toxicity is further based on the one or more voice emotion parameters.
19. The non-transitory computer-readable medium of claim 15 , wherein the audio stream is part of a video stream, and the operations further comprise:
analyzing the audio stream to identify an instance of toxicity;
detecting a portion of video frames in the video stream that depict an avatar associated with the first user performing an action that corresponds to the instance of toxicity in the audio stream; and
responsive to detecting the portion of video frames in the video stream that depict the action, modifying at least the portion of the video frames in the video stream by one or more of blurring the portion or replacing the portion with pixels that match a background region.
20. The non-transitory computer-readable medium of claim 19 , wherein the action is mouth movement of the avatar that forms words to correspond to the instance of toxicity in the audio stream.