Audio-based link generation

1. A system comprising a computer comprising a processor and a memory, the memory storing instructions executable by the processor such that the computer is programmed to:

determine that first speech data received from a first user via a first device and second speech data received from a second user via a second device are from a dialog occurring in real time between the first user and the second user;

receive a plurality of first audio segments of the dialog from the first user via the first device, and a plurality of second audio segments of the dialog from the second user via the second user device, the first audio segments and the second audio segments associated with respective timestamps;

stitch together the first audio segments and the second audio segments based on the respective timestamps to generate a stitched audio file representing at least a portion of the dialog;

compute a correlation index, based on the stitched audio file, that provides a measure of whether content of the dialog is likely to influence future behavior of the first user and/or the second user; and

generate, responsive to determining that the correlation index exceeds a correlation threshold, a link comprising a set of digital data that specifies an association between first user and the second user.

2. The system of claim 1 , wherein:

each of the first audio segments and the second audio segments is associated with at least a respective starting timestamp and a respective ending timestamp; and

the stitching together comprises concatenating the first audio segments and the second audio segments into a sequence of audio segments based on identifying matches between the respective starting timestamps and the respective ending timestamps.

3. The system of claim 2 , wherein:

the identifying matches comprises determining whether any of the respective starting timestamps of a first of the first audio segments and/or the second audio segments is within a predetermined temporal range of any of the respective ending timestamps of a second of the first audio segments and/or the second audio segments.

4. The system of claim 1 , wherein the generating the correlation index is further based on:

prompting the first user via the first device, responsive to the determining that the first speech data received from the first user via the first device and the second speech data received from the second user via the second device are from the dialog, for a rating indicating a predicted benefit to the first user and/or the second user from generating the link based on the dialog; and

receiving the rating from the first user via the first device responsive to the prompting.

5. The system of claim 1 , wherein the generating the correlation index comprises evaluating the stitched audio file against a predefined rule that automatically sets the correlation index to below the correlation threshold upon detecting any one or more of a predefined set of conditions, such that the generating the link is not performed upon the detecting.

6. The system of claim 1 , wherein the generating the correlation index comprises quantitatively evaluating a quality of interaction between the first user and the second user based on determining a length of the dialog.

7. The system of claim 1 , wherein the generating the correlation index comprises quantitatively evaluating a quality of interaction between the first user and the second user based on parsing the stitched audio file to identify positive and/or negative language content of the dialog.

8. The system of claim 1 , wherein the generating the correlation index comprises quantitatively evaluating a quality of interaction between the first user and the second user based on applying computational voice analysis to the stitched audio file to assign emotional scores to the first user and/or the second user during the dialog.

9. The system of claim 1 , wherein the generating the correlation index comprises scoring a difference in affinity and/or keywords between the first user and the second user.

10. The system of claim 1 , wherein the generating the correlation index comprises:

generating a first partial correlation index based on indicators of a quality of interaction between the first user and the second user;

generating a second partial correlation index based on scoring a difference in affinity between the first user and the second user and/or scoring a change in keywords between the first user and the second user; and

computing the correlation index as a function of the first partial correlation index and the second partial correlation index.

11. The system of claim 1 , wherein the generating the correlation index comprises training a neural network to output the correlation index based on inputting at least the stitched audio file.

12. The system of claim 1 , wherein the generating the link comprises:

directing the first device and/or the second device to request user input as to whether to generate the link; and

generating the link only upon receipt of the user input from the first device and/or the second device indicating confirmation to generate the link.

13. The system of claim 1 , wherein the determining that the first speech data received from the first user via the first device and the second speech data received from the second user via the second device are from the dialog comprises:

determining that the respective timestamps fall within a time window during which the first device is determined to be within a predetermined threshold distance of the second device.

14. The system of claim 1 , wherein the generating the link comprises:

determining, based on analyzing the stitched audio file, an object that is a subject of the association between the first user and the second user; and

generating the set of digital data to further specify the object.

15. The system of claim 14 , further comprising:

outputting the link to at least the first device as a selectable link by which to obtain information about the object.

16. A method comprising:

determining that first speech data received from a first user via a first device and second speech data received from a second user via a second device are from a dialog occurring in real time between the first user and the second user;

receiving a plurality of first audio segments of the dialog from the first user via the first device, and a plurality of second audio segments of the dialog from the second user via the second user device, the first audio segments and the second audio segments associated with respective timestamps;

stitching together the first audio segments and the second audio segments based on the respective timestamps to generate a stitched audio file representing at least a portion of the dialog;

computing a correlation index, based on the stitched audio file, that provides a measure of whether content of the dialog is likely to influence future behavior of the first user and/or the second user; and

generating, responsive to determining that the correlation index exceeds a correlation threshold, a link comprising a set of digital data that specifies an association between first user and the second user.

17. The method of claim 16 , wherein the computing the correlation index comprises evaluating the stitched audio file against a predefined rule that automatically sets the correlation index to below the correlation threshold upon detecting any one or more of a predefined set of conditions, such that the generating the link is not performed upon the detecting.

18. The method of claim 16 , wherein the computing the correlation index comprises quantitatively evaluating a quality of interaction between the first user and the second user based on: determining a length of the dialog; parsing the stitched audio file to identify positive and/or negative language content of the dialog; or applying computational voice analysis to the stitched audio file to assign emotional scores to the first user and/or the second user during the dialog.

19. The method of claim 16 , wherein the computing the correlation index comprises: generating a first partial correlation index based on indicators of a quality of interaction between the first user and the second user;

generating a second partial correlation index based on scoring a difference in affinity between the first user and the second user and/or scoring a change in keywords between the first user and the second user; and computing the correlation index as a function of the first partial correlation index and the second partial correlation index.

20. The method of claim 16 , wherein the computing the correlation index comprises training a neural network to output the correlation index based on inputting at least the stitched audio file.