Task completion based on speech analysis

1. A system comprising:

a processor; and

a memory component having instructions stored thereon, when executed by the processor, causes the processor to perform operations comprising:

receiving an audio signal of a communication session between a member-related client device and an agent client device, wherein a caller is associated with the member-related client device;

processing the audio signal to generate an audio caller utterance and a transcribed caller utterance;

generating an identified task based on the transcribed caller utterance;

sampling the audio caller utterance to generate a plurality of samples of the audio caller utterance;

generating a loudness result based on loudness values of the plurality of samples using a loudness neural network associated with the identified task;

generating a pitch result based on pitch values of the plurality of samples using a pitch neural network associated with the identified task;

generating a tone result for a plurality of words in the transcribed caller utterance using a tone neural network associated with the identified task; and

generating a task completion probability result using a task completion probability neural network associated with the identified task, wherein generating the task completion probability result is based on the loudness result, the pitch result, or the tone result.

2. The system of claim 1 , wherein the processor to perform operations further comprising:

determining whether a member identification is associated with the audio signal;

when the member identification is associated with the audio signal,

obtaining a member context associated with the member identification,

computing a criticality value that indicates a level of criticality associated with the communication session based on the member context; and

when there is no member identification associated with the audio signal,

setting the criticality value to indicate a low criticality.

3. The system of claim 2 , wherein generating the task completion probability result is further based on the criticality value.

4. The system of claim 1 , wherein the processor to perform operations further comprising:

generating a relationship data based on the transcribed caller utterance.

5. The system of claim 4 , wherein generating the task completion probability result is further based on the relationship data.

6. The system of claim 1 , wherein processing the audio signal to generate the audio caller utterance and the transcribed caller utterance further comprises:

processing the audio signal using a speech-to-text engine

to transcribe the audio signal into a transcribed audio signal,

to identify separate parties in the audio signal, wherein the separate parties include the caller and an agent, and

to provide start and end times for each of a plurality of utterances included in the audio signal;

removing agent utterances from the plurality of utterances included in the audio signal;

separating the audio caller utterance from the audio signal using the start and end times for each of a plurality of utterances; and

selecting the transcribed caller utterance from the transcribed audio signal based on the audio caller utterance.

7. The system of claim 1 , wherein the processor to perform operations further comprising:

generating a relationship data based on the transcribed caller utterance, wherein the relationship data includes at least one potential relationship between the caller and a patient that is the subject of the transcribed caller utterance.

8. The system of claim 1 , wherein the system further comprises:

a plurality of neural networks including a plurality of loudness neural networks, a plurality of pitch neural networks, a plurality of tone neural networks, and a plurality of task completion probability neural networks, wherein the plurality of neural networks are associated with a plurality of tasks, respectively.

9. The system of claim 1 , wherein generating the loudness result based on loudness values of the plurality of samples using the loudness neural network associated with the identified task includes:

computing the plurality of loudness values for the plurality of samples;

normalizing the loudness values to obtain normalized loudness values; and

generating the loudness result based on the normalized loudness values.

10. The system of claim 1 , wherein generating the pitch result based on pitch values of the plurality of samples using the pitch neural network associated with the identified task includes:

computing the plurality of pitch values for the plurality of samples;

normalizing the pitch values to obtain normalized pitch values; and

generating the pitch result based on the normalized pitch values.

11. The system of claim 1 , wherein the communication session includes at least one of: an interactive voice response (IVR) or a voice call.

12. A method comprising:

receiving an audio signal of a communication session between a member-related client device and an agent client device, wherein a caller is associated with the member-related client device;

processing the audio signal to generate an audio caller utterance and a transcribed caller utterance;

generating an identified task based on the transcribed caller utterance;

sampling the audio caller utterance to generate a plurality of samples of the audio caller utterance;

generating a loudness result based on loudness values of the plurality of samples using a Long Short-Term Memory (LTSM) loudness neural network associated with the identified task;

generating a pitch result based on pitch values of the plurality of samples using a Long Short-Term Memory (LTSM) pitch neural network associated with the identified task;

generating a Long Short-Term Memory (LTSM) tone result for a plurality of words in the transcribed caller utterance using a tone neural network associated with the identified task; and

generating a task completion probability result using a task completion probability convolutional neural network (CNN) associated with the identified task, wherein generating the task completion probability result is based on the loudness result, the pitch result, or the tone result.

13. The method of claim 12 , further comprising:

determining whether a member identification is associated with the audio signal;

when the member identification is associated with the audio signal,

obtaining a member context associated with the member identification,

computing a criticality value that indicates a level of criticality associated with the communication session based on the member context; and

when there is no member identification associated with the audio signal,

setting the criticality value to indicate a low criticality.

14. The method of claim 13 , wherein generating the task completion probability result is further based on the criticality value.

15. The method of claim 12 , further comprising:

generating a relationship data based on the transcribed caller utterance.

16. The method of claim 15 , wherein generating the task completion probability result is further based on the relationship data.

17. The method of claim 12 , wherein processing the audio signal to generate the audio caller utterance and the transcribed caller utterance further comprises:

processing the audio signal using a speech-to-text engine

to transcribe the audio signal into a transcribed audio signal,

to identify separate parties in the audio signal, wherein the separate parties include the caller and an agent, and

to provide start and end times for each of a plurality of utterances included in the audio signal;

removing agent utterances from the plurality of utterances included in the audio signal;

separating the audio caller utterance from the audio signal using the start and end times for each of a plurality of utterances; and

selecting the transcribed caller utterance from the transcribed audio signal based on the audio caller utterance.

18. The method of claim 12 , further comprising:

generating a relationship data based on the transcribed caller utterance, wherein the relationship data includes at least one potential relationship between the caller and a patient that is the subject of the transcribed caller utterance.

19. The method of claim 12 , wherein generating the loudness result based on loudness values of the plurality of samples using the loudness neural network associated with the identified task includes:

computing the plurality of loudness values for the plurality of samples;

normalizing the loudness values to obtain normalized loudness values; and

generating the loudness result based on the normalized loudness values.

20. The method of claim 12 , wherein generating the pitch result based on pitch values of the plurality of samples using the pitch neural network associated with the identified task includes:

computing the plurality of pitch values for the plurality of samples;

normalizing the pitch values to obtain normalized pitch values; and

generating the pitch result based on the normalized pitch values.

21. A non-transitory storage medium having instructions stored thereon, when executed by a processor causes the processor to perform operations comprising:

processing the audio signal to generate an audio caller utterance and a transcribed caller utterance;

generating an identified task based on the transcribed caller utterance;

sampling the audio caller utterance to generate a plurality of samples of the audio caller utterance;

generating a loudness result based on loudness values of the plurality of samples using a loudness neural network associated with the identified task;

generating a pitch result based on pitch values of the plurality of samples using a pitch neural network associated with the identified task;

generating a tone result for a plurality of words in the transcribed caller utterance using a tone neural network associated with the identified task; and

generating a task completion probability result using a task completion probability neural network associated with the identified task, wherein generating the task completion probability result is based on the loudness result, the pitch result, or the tone result.