Abstract: A computer-implemented method of predicting purpose of call for a call center is disclosed to support various analytical functions, Automatic discovery of call topics, clustering of topics, and assigning of utterance to clusters may be provided. In some implementations, a pipeline includes an inference engine that includes a machine learning model trained on call transcripts and call attribute data to predict purpose of call.

Abstract: Functionality is provided for the automated creation, testing, training, adaptation and deployment of AI models and changes thereto. Base classes are provided that enable practicable creation of new models from existing one. New models are tested on live data sets offline from user sites. New training methods are provided for the production of particular outcomes. Efficient adaptation of new AI models is facilitated, encompassing data scientist and development team control over how fast to train and deploy new models.

Abstract: A system and method are described for automatic acoustic feedback cancellation in real time. In some implementations, the system may receive audio data describing an audio signal, which the system may use to determine a set of frames of the audio signal. Spectral analysis may be performed on the one or more frames of the audio to detect spectral patterns of two or more frames indicative of acoustic feedback. An additional delay identification test may be performed to identify a consistent delay indicative of acoustic feedback. In some implementations, a state machine is advanced based in part on accumulated delay votes. Decisions can be made to mute the acoustic feedback and cease the muting operation when silence is detected.

Abstract: Audio data saved at the end of client interactions are sampled, analyzed for pauses in speech, and sliced into stretches of acoustic data containing human speech between those pauses. The acoustic data are accompanied by machine transcripts made by VoiceAI. A suitable distribution of data useful for training and testing are stipulated during data sampling by applying certain filtering criteria. The resulting datasets are sent for transcription by a human transcriber team. The human transcripts are retrieved, some post-transcription processing and cleaning are performed, and the results are added to datastores for training and testing an acoustic model.

Abstract: A system and computer-implemented method for providing telephony communication services for VoIP or analog telephony devices using an on-premises gateway and remotely located VoIP system are described. Some implementations of the methods may include generating a configuration file mapping a plurality of tag extensions to a plurality of analog telephone ports of a high-density analog telephony adapter (HDATA), which may be communicatively coupled to the VoIP system. The VoIP system may transmit the configuration file to the HDATA, receive a telephony service request, and determine a tag extension based on the telephony service request. The VoIP system may transmit an identification of the tag extension to the HDATA and route a telephony service to the HDATA via the communication channel using the tag extension.

Abstract: A proactive autoscaling system can use heuristics and machine learning to proactively, dynamically and automatically scale computing resources allocated to applications up and down, prior to scaling events that cause changes in load levels. The proactive autoscaling system may be stateless, and may be agnostic to use case, platform, field of endeavor, or communication protocol used by the applications. The proactive autoscaling system gleans metrics in real-time. The gleaned metrics are indicative of load levels concerning one or more applications. These gleaned metrics may be in a variety of formats, and may be from different sources, both internal or external to the applications. The proactive autoscaling system automatically predicts scaling events based on gleaned metrics. Prior to the occurrence of a predicted scaling event, the proactive autoscaling system can automatically scale computing resources available to one or more target applications, in response to the predicting of the scaling event.

Abstract: Functionality is provided for the automated creation, testing, training, adaptation and deployment of AI models and changes thereto. Base classes are provided that enable practicable creation of new models from existing one. New models are tested on live data sets offline from user sites. New training methods are provided for the production of particular outcomes. Efficient adaptation of new AI models is facilitated, encompassing data scientist and development team control over how fast to train and deploy new models.

Abstract: A system and method are described for smart noise cancellation. In some implementations, the system may receive audio data describing an audio signal, which the system may use to determine a set of frames of the audio signal. Spectral analysis, which may include a signal-to-noise ratio estimate, may be performed on the one or more frames of the audio. In some instances, the system may identify a noise frame from among the one or more frames based on the spectral analysis, and may reduce noise in the one or more frames based on the noise frame and the spectral analysis on the one or more frames of the audio signal.

Abstract: The content and/or sound characteristics of audio calls are monitored in real-time. The occurrence of triggers during audio calls is automatically detected, and in response directives are displayed to parties in real-time while calls are occurring. Multiple triggers and directives can be defined and edited, either by human users and/or automatically (e.g., in response to results obtained and tracked over time). Each given directive corresponds to one or more given triggers, such that detection of an occurrence of a given trigger during a monitored audio call results in the automatic outputting of a corresponding directive.

Abstract: Audio data describing an audio signal may be received and used to determine a set of frames of the audio signal. One or more potential music events may be determined in the audio signal using a spectral analysis of the set of frames. The audio signal may be analyzed for one or more potential noise or tone events. One or more music states of the audio signal may be determined based on the one or more potential music events and a presence or absence of the one or more noise or tone events. Audio enhancement of the audio signal may be modified based on the one or more determined states of the audio signal.

Abstract: An AI platform is fully integrated with existing VoIP/AVoIP telephony infrastructure. In the course of providing VoIP/AVoIP audiovisual calls, a VoIP/AVoIP media stream of audiovisual content is processed, and transferred between endpoints. AI processing is applied to each frame of the transferred audiovisual content, in real-time while the audiovisual call is occurring. For example, automated speech recognition can be performed on the content, in which the speech of the audiovisual content is converted to text. The audiovisual call can further be automatically transcribed to a text file in real-time. Another example is the automatic detection of the occurrence of specific triggers during calls. Additional enhanced functionality is automatically provided as a result of applying the AI processing to the transferred audiovisual content. For example, in response to detecting the occurrence of a specific trigger, a corresponding directive can be automatically output on a screen of a calling device.

Abstract: On premises gateways located within organization sites interconnect VoIP systems, the public switched telephone network (PSTN), Private Branch Exchanges and other telephony infrastructure.

Abstract: Transport layer protocol packet headers are used to encode application layer attributes in the context of an AVoIP platform. An endpoint encodes signaling information in the transport layer protocol packet header of an audiovisual stream, (e.g., in the synchronization source identifier (“SSRC”) field of an RTP header). The signaling information may include requests to add or remove the audiovisual stream to/from an existing videoconference, an application layer identifier, and metadata concerning audiovisual content contained in the audiovisual stream such as the resolution, codec, etc. After adding the signaling information, the endpoint transmits the audiovisual stream to a server.

Abstract: Audio data describing an audio signal may be received and used to determine a set of frames of the audio signal. One or more potential music events may be determined in the audio signal using a spectral analysis of the set of frames. The audio signal may be analyzed for one or more potential noise or tone events. One or more music states of the audio signal may be determined based on the one or more potential music events and a presence or absence of the one or more noise or tone events. Audio enhancement of the audio signal may be modified based on the one or more determined states of the audio signal.

Abstract: Audio data describing an audio signal may be received and used to determine a set of frames of the audio signal. A plurality of note onsets in the set of frames may be identified based on spectral energy of the audio signal in the set of frames. One or more tempos may be computed based on the identified plurality of note onsets. The one or more tempos may be validated based on a tempo validation condition. One or more music states of the audio signal may be determined based on the validated one or more tempos. Audio enhancement of the audio signal may be modified based on the one or more determined states of the audio signal.