Multi-party conversation analyzer and logger
In one aspect, the present invention facilitates the investigation of networks of criminals, by gathering associations between phone numbers, the names of persons reached at those phone numbers, and voice print data. In another aspect the invention automatically detects phone calls from a prison where the voiceprint of the person called matches the voiceprint of a past inmate. In another aspect the invention detects identity scams in prisons, by monitoring for known voice characteristics of likely imposters on phone calls made by prisoners. In another aspect, the invention automatically does speech-to-text conversion of phone numbers spoken within a predetermined time of detecting data indicative of a three-way call event while monitoring a phone call from a prison inmate. In another aspect, the invention automatically thwarts attempts of prison inmates to use re-dialing services. In another aspect, the invention automatically tags audio data retrieved from a database, by steganographically encoding into the audio data the identity of the official retrieving the audio data.
1. A method for determining, by a communications system providing monitored communication services to residents of a controlled-environment facility, an impersonation of a first person in a monitored conversation, the method comprising:
Monitoring a voice conversation conducted using a communication device, wherein a participant in the monitored voice conversation is a person purporting to be the first person;
determining a set of candidate voice models associated with possible imposters of the first person, wherein each candidate voice model is determined based in part on a similarity of voice model of a candidate and a voice model of the first person; periodically comparing one or more of the candidate voice models against the voice of the person purporting to be the first person;
ranking the set of candidate voice models based on the likelihood an individual associated with each candidate voice model is posing as the person purporting to be the first person;
detecting an impersonation of the first person in the monitored conversation, wherein the impersonation is detected by identifying variations above a specified threshold between voice characteristics of the purported first person in the monitored conversation and the voice model of the first person, wherein the specified threshold is determined, based in part on the size of the set of the determined candidate voice models; wherein the variations are determined based on a multi-dimensional distance between the voice characteristics of the purported first person in the monitored conversation and the corresponding voice characteristics of the voice model of the first person; and
Signaling, by the communications system, the detection of the impersonation of the first person;
Displaying, on a computer display, the ranked set of candidate voice models, and, in response to the signal, a notification of the detected impersonation.
2. The method of claim 1 , wherein the possible imposters of the first person include persons known to have previously used a false identity.
3. The method of claim 1 , further comprising:
wherein the displayed voice models are sorted with the mostly likely imposter of the first person is displayed first in the listing.
4. The method of claim 1 , wherein the possible imposters of the first person include persons determined to have previously participated in a phone call using the phone number being used for the monitored conversation.
5. The method of claim 1 , wherein the possible imposters of the first person include persons with access to the communication device.
6. The method of claim 1 , wherein the monitoring comprises determining one or more groupings of voice characteristics for a voice participating in the monitored conversation.
7. The method of claim 6 , wherein the possible imposters of the first person include persons with voice characteristics that fall within the one or more determined groupings.
8. The method of claim 1 wherein the specified threshold is also determined based on the similarity of a voice model of the first person to the candidate voice model of a most likely imposter of the first person.
9. The method of claim 1 , wherein the possible imposters of the first person are determined based on the multi-dimensional distance between a plurality of voice characteristics of the voice model of the first person and the corresponding plurality of voice characteristics in the individual candidate voice models.
10. The method of claim 9 , wherein the multi-dimensional distance comprises one or more of a drift distance and a drift direction.
11. A non-transitory computer-accessible storage medium storing program instructions that, when executed by a data processing device, cause the data processing device to implement operations for determining an impersonation of a first person in a monitored conversation, the operations comprising:
monitoring a voice conversation conducted on a communication device, wherein a participant in the monitored voice conversation is a person purporting to be a first person;
determining a set of candidate voice models associated with possible imposters of the first person, wherein each candidate voice model is determined based in part on a similarity of the voice model of a candidate and a voice model of the first person;
Periodically comparing one or more of the candidate voice models against the voice of the possible imposter participating in the monitored conversation; and
ranking the set of candidate voice models based on the likelihood an individual associated with each candidate voice model is posing as the first person in the monitored voice conversation; and
detecting an impersonation of the first person in the monitored conversation, wherein the impersonation is detected by identifying variations above a specified threshold between voice characteristics of the purported first person in the monitored conversation and the voice model of the first person, wherein the variations are determined based on a multi-dimensional distance between the voice characteristics of the purported first person in the monitored conversation and the corresponding voice characteristics of the voice model of the first person, and wherein the specified threshold is determined, based in part on the size of the set of the determined candidate voice models; and
Signaling, by the data processing device, the detection of the impersonation of the first person;
Displaying, on a computer display, the ranked set of candidate voice models, and, in response to the signal, a notification of the detected impersonation.
12. The non-transitory computer-accessible storage medium of claim 11 , wherein the possible imposters of the first person include persons known to have previously used a false identity.
13. The non-transitory computer-accessible storage medium of claim 11 , the operations further comprising: wherein the displayed voice models are sorted with the mostly likely imposter of the first person is displayed first in the listing.
14. The non-transitory computer-accessible storage medium of claim 11 , wherein the possible imposters of the first person include persons determined to have previously participated in a phone call using the phone number being used for the monitored conversation.
15. The non-transitory computer-accessible storage medium of claim 11 , wherein the possible imposters of the first person include persons with access to the communication device.
16. The non-transitory computer-accessible storage medium of claim 11 , wherein the monitoring comprises determining one or more groupings of voice characteristics for a voice participating in the monitored conversation.
17. The non-transitory computer-accessible storage medium of claim 16 , wherein the possible imposters of the first person include persons with voice characteristics that fall within the one or more determined groupings.
18. The non-transitory computer-accessible storage medium of claim 11 , wherein the specified threshold is also determined based on the similarity of a voice model of the first person to the candidate voice model of a most likely imposter of the first person.
19. The non-transitory computer-accessible storage medium of claim 11 , wherein the possible imposters of the first person are determined based on a multi-dimensional distance between a plurality of voice characteristics of the voice models model of the first person and the corresponding plurality of voice characteristics in the individual candidate voice models.
20. The non-transitory computer-accessible storage medium of claim 19 , wherein the multi-dimensional distance comprises one or more of a drift distance and a drift direction.