Speech analysis models as a service
A model as a service system in which call participants' emotional or mental state is recognized by the system based on aspects of voice-based audio data collected from the call participants' communication devices, in particular the state of a company's customers. The system provides an algorithm for use by a company employee during an ongoing communication with a customer of the company. The voice-based data is recorded from the participants' communication devices, and sent to a server that analyzes the recordings for characteristics that reflect the current emotional or mental state of call participants, particularly the state of the customer. The characteristics are used to generate an algorithm that provides to a company participant suggestions for modifying aspects of their voice communication with the customer in real- or near real-time.
1. A model-as-a-service system comprising:
a server including:
a subscription module;
a subscription database;
a data collection module;
a source data base;
a correlation module;
an algorithm creation module; and
an algorithm database;
a data communication network communicatively coupled to the server at a first network location; and
a user device communicatively coupled to the network at a second network location,
wherein a user device base module is configured to:
communicatively connect to the subscription module;
send a user ID to the subscription module;
in the case the user is not registered, send user device information to the subscription module for registration;
send important business aspects of the user to the subscription module;
repeatedly poll for a request, from the data collection module, for the user device to send a user device audio database, and receive the request; and
send the user device audio database to the data collection module.
2. The system of claim 1 , wherein the subscription module is configured to:
poll the user device for connection information;
communicatively connect to the user device;
receive a user ID from the user device;
compare the received user ID to the subscription database and determine whether the user ID matches an entry in the subscription database;
in the case the user ID does not match an entry in the subscription database, register the user device, including:
storing the user ID in the subscription database,
receiving important business aspects from the user device,
storing the received business aspects in the subscription database; and
initiating the data collection module.
3. The system of claim 1 , wherein the data collection module is configured to:
send a request to the user device base module for an audio database of the user device;
receive the requested audio database from the user device;
analyze audio files stored in the received audio database;
store data pertaining to the analyzed audio files in the source database; and
initiate the correlation module.
4. The system of claim 1 , wherein the correlation module is configured to:
filter the source database for parameters of data stored in the source database;
send the parameters to the algorithm creation module, including:
select a first parameter of the stored information,
correlate the first parameter data with data of additional parameters of the stored information,
determine whether there is a correlation greater than 95% between data of the first parameter and data of any of the additional parameters,
in the case there is a correlation greater than 95%, extract a most recurring datapoint of the first parameter for each of the additional parameters,
send information of the most recurring datapoint to the algorithm creation module,
determine whether there is a non-correlated parameter of the stored information;
in the case there is a non-correlated parameter, select the non-correlated parameter as a new first parameter, and
return to the step of correlating first parameter data with data of additional parameters of the stored information, and proceed from there until there are no non-correlated parameters remaining; and
initiate the algorithm creation module.
5. The system of claim 1 , wherein the algorithm creation module is configured to:
receive correlation data points from the correlation module;
filter the subscription database for subscriber information of the user;
extract from the subscription database important business aspects of the user;
filter the algorithm database on the extracted business aspects;
select an unselected one of the received data points;
filter the algorithm database on parameters of the selected data point;
search the filtered algorithm database for algorithms corresponding to the data point;
in the case there is a match, extract the corresponding algorithm;
in the case there is another unselected received datapoint, select the unselected received datapoint as a new selected data point and return to the step of filtering the algorithm database on parameters of the new selected datapoint, and proceed from there; and
combine the extracted algorithms as a created algorithm for use in real- or near real-time information of the user device.
6. A model-as-a-service method comprising:
obtaining a server including:
a subscription module,
a subscription database,
a data collection module,
a source data base,
a correlation module,
an algorithm creation module, and
an algorithm database;
communicatively coupling the server to a data communication network at a first network location; and
communicatively coupling a user device to the network at a second network location,
wherein a user device base module performs the steps comprising:
communicatively connecting to the subscription module;
sending a user ID to the subscription module;
in the case the user device is not registered, sending information of the user device to the subscription module;
sending, to the subscription module, important business aspects of the user device;
repeatedly polling for a request, from the data collection module, for the user device to send a user device audio database, and receiving the request; and
sending the user device audio data base to the data collection module.
7. The method of claim 6 , wherein the subscription module performs the steps comprising:
polling the user device for connection information;
communicatively connecting to the user device;
receiving a user ID from the user device;
comparing the received user ID to the subscription database and determining whether the user ID matches an entry in the subscription database;
in the case the user ID does not match an entry in the subscription database, registering the user device, including:
storing the user ID in the subscription database,
receiving important business aspects from the user device,
storing the received business aspects in the subscription database; and
initiating the data collection module.
8. The method of claim 6 , wherein the data collection module performs the steps comprising:
sending a request to the user device base module for an audio database of the user device;
receiving the requested audio database from the user device;
analyzing audio files stored in the received audio database;
storing data pertaining to the analyzed audio files in the source database; and
initiating the correlation module.
9. The method of claim 6 , wherein the correlation module performs the steps comprising:
filtering the source database for data of parameters of a company;
sending the company parameters data to the algorithm creation module, including:
selecting a first parameter of the company information,
correlating data of the first parameter with data of additional parameters of the company information,
determining whether there is a correlation greater than 95% between the first parameter and any of the additional parameters,
in the case there is a correlation greater than 95%, extracting a most recurring datapoint of the first parameter for each of the additional parameters,
sending information of the most recurring datapoint to the algorithm creation module,
determining whether there is a non-correlated parameter remaining,
in the case there is a non-correlated parameter remaining, selecting the non-correlated parameter as a new first parameter,
returning to the step of correlating data of the first parameter with data of additional parameters using the new first parameter; and
when there are no remaining non-correlated parameters remaining, initiating the algorithm creation module.
10. The method of claim 6 , wherein the algorithm creation module is configured to perform the steps comprising:
receiving, from a correlation module, correlation data points and corresponding company information;
filtering a subscription database for information on the company;
extracting from the subscription database important business aspects of the company;
filtering an algorithm database on the extracted business aspects;
selecting an unselected one of the received data points;
filtering the algorithm database on parameters of the selected data point;
searching the filtered algorithm database for algorithms corresponding to the data point;
in the case there is a match, extracting the corresponding algorithm;
in the case there is another unselected received datapoint, returning to the step of selecting an unselected one of the received data points; and
combining the extracted algorithms as a created algorithm for use in real or near real time on information of the user device.
11. A model-as-a-service system comprising:
a server including, a subscription module, a subscription database, a data collection module, a source data base, a correlation module, an algorithm creation module, and an algorithm database;
a data communication network communicatively coupled to the server at a first network location; and
a user device communicatively coupled to the network at a second network location,
wherein the correlation module is configured to:
filter the source database for parameters of data stored in the source database;
send the parameters to the algorithm creation module, including:
select a first parameter of the stored information,
correlate the first parameter data with data of additional parameters of the stored information,
determine whether there is a correlation greater than 95% between data of the first parameter and data of any of the additional parameters,
in the case there is a correlation greater than 95%, extract a most recurring datapoint of the first parameter for each of the additional parameters,
send information of the most recurring datapoint to the algorithm creation module,
determine whether there is a non-correlated parameter of the stored information;
in the case there is a non-correlated parameter, select the non-correlated parameter as a new first parameter, and
return to the step of correlating first parameter data with data of additional parameters of the stored information, and proceed from there until there are no non-correlated parameters remaining; and
initiate the algorithm creation module.
12. The system of claim 11 , wherein the subscription module is configured to:
poll the user device for connection information;
communicatively connect to the user device;
receive a user ID from the user device;
compare the received user ID to the subscription database and determine whether the user ID matches an entry in the subscription database;
in the case the user ID does not match an entry in the subscription database, register the user device, including:
storing the user ID in the subscription database,
receiving important business aspects from the user device,
storing the received business aspects in the subscription database; and
initiating the data collection module.
13. A model-as-a-service system comprising:
a server including a subscription module, a subscription database, a data collection module, a source data base, a correlation module, an algorithm creation module, and an algorithm database;
a data communication network communicatively coupled to the server at a first network location; and
a user device communicatively coupled to the network at a second network location,
wherein the algorithm creation module is configured to:
receive correlation data points from the correlation module;
filter the subscription database for subscriber information of the user;
extract from the subscription database important business aspects of the user;
filter the algorithm database on the extracted business aspects;
select an unselected one of the received data points;
filter the algorithm database on parameters of the selected data point;
search the filtered algorithm database for algorithms corresponding to the data point;
in the case there is a match, extract the corresponding algorithm;
in the case there is another unselected received datapoint, select the unselected received datapoint as a new selected data point and return to the step of filtering the algorithm database on parameters of the new selected datapoint, and proceed from there; and
combine the extracted algorithms as a created algorithm for use in real- or near real-time information of the user device.
14. The system of claim 13 , wherein the data collection module is configured to:
send a request to a user device base module for an audio database of the user device;
receive the requested audio database from the user device;
analyze audio files stored in the received audio database;
store data pertaining to the analyzed audio files in the source database; and
initiate the correlation module.
15. A model-as-a-service method comprising:
obtaining a server including:
a subscription module,
a subscription database,
a data collection module,
a source data base,
a correlation module,
an algorithm creation module, and
an algorithm database;
communicatively coupling the server to a data communication network at a first network location; and
communicatively coupling a user device to the network at a second network location,
wherein the correlation module performs the steps comprising:
filtering the source database for data of parameters of the company;
sending a company parameters data to the algorithm creation module, including:
selecting a first parameter of the company information,
correlating data of the first parameter with data of additional parameters of the company information,
determining whether there is a correlation greater than 95% between the first parameter and any of the additional parameters,
in the case there is a correlation greater than 95%, extracting a most recurring datapoint of the first parameter for each of the additional parameters,
sending information of the most recurring datapoint to the algorithm creation module,
determining whether there is a non-correlated parameter remaining,
in the case there is a non-correlated parameter remaining, selecting the non-correlated parameter as a new first parameter,
returning to the step of correlating data of the first parameter with data of additional parameters using the new first parameter; and
when there are no remaining non-correlated parameters remaining, initiating the algorithm creation module.
16. The method of claim 15 , wherein the subscription module performs the steps comprising:
polling the user device for connection information;
communicatively connecting to the user device;
receiving a user ID from the user device;
comparing the received user ID to the subscription database and determining whether the user ID matches an entry in the subscription database;
in the case the user ID does not match an entry in the subscription database, registering the user device, including:
storing the user ID in the subscription database,
receiving important business aspects from the user device,
storing the received business aspects in the subscription database; and
initiating the data collection module.
17. A model-as-a-service method comprising:
obtaining a server including:
a subscription module,
a subscription database,
a data collection module,
a source data base,
a correlation module,
an algorithm creation module, and
an algorithm database;
communicatively coupling the server to a data communication network at a first network location; and
communicatively coupling a user device to the network at a second network location,
wherein the algorithm creation module is configured to perform the steps comprising:
receiving, from a correlation module, correlation data points and corresponding company information;
filtering a subscription database for information on the company;
extracting from the subscription database important business aspects of the company;
filtering an algorithm database on the extracted business aspects;
selecting an unselected one of the received data points;
filtering the algorithm database on parameters of the selected data point;
searching the filtered algorithm database for algorithms corresponding to the data point;
in the case there is a match, extracting the corresponding algorithm;
in the case there is another unselected received datapoint, returning to the step of selecting an unselected one of the received data points; and
combining the extracted algorithms as a created algorithm for use in real or near real time on information of the user device.
18. The method of claim 17 , wherein the data collection module performs the steps comprising:
sending a request to a user device base module for an audio database of the user device;
receiving the requested audio database from the user device;
analyzing audio files stored in the received audio database;
storing data pertaining to the analyzed audio files in the source database; and
initiating the correlation module.