Identical conversation detection method and apparatus
View Patent ↗An automated system is disclosed for detecting situations in which identical segments of conversation appear within two different recordings. The system automatically detects where within each of two audio recordings an identical conversation segment begins and ends, thus enabling detection of conversations where multiple prison inmates participated in a conference call.
1. A system for detecting disallowed voice communications between a first person and second person, the system comprising:
one or more processors; and
a memory device coupled to the one or more processors, the memory device storing computer-readable instructions that, upon execution by the one or more processors, cause the system to:
record digital audio captured at a first communication device used by the first person to produce a first recording of the first person participating in a first communication;
record digital audio captured at a second communication device used by the second person to produce a second recording of the second person participating in a second communication;
generate a first reduced data set from the first recording, wherein the first reduced data comprises a first voiceprint;
generate a second reduced data set form the second recording, wherein the second reduced data comprises a second voiceprint; and
correlate the first reduced data set and the second reduced data set to in order to identify portions of the first recording and the second recording that overlap in time, wherein the overlapping portions are correlated greater than a pre-determined threshold, and wherein the correlation of the overlapping portions indicates a disallowed voice communication between the first person and the second person.
2. The system of claim 1 , wherein the first reduced data set comprises a first signature vector and the second reduced data set comprises a second signature vector.
3. The system of claim 2 , further comprising
detecting a first recording-point-invariant audio event in the first voice recording and a second recording-point-invariant audio event in the second voice recording.
4. The system of claim 3 , wherein the first signature vector is calculated based on the first recording-point-invariant audio event and the second signature vector is calculated based on the second recording-point-invariant audio event.
5. The system of claim 4 , wherein the first recording-point-invariant audio event and the second recording-point-invariant audio event have different amplitudes.
6. The system of claim 5 , further comprising:
correlating the first signature vector and the second signature vector in order to determine that the first voice recording and the second voice recording comprise a voice communication between the first person and the second person.
7. The system of claim 5 , wherein the first recording-point-invariant audio event and the second recording-point-invariant audio event each comprise one or more vowel sounds.
8. The system of claim 1 , wherein the first communication comprises a communication between the first person and a third person and wherein the second communication comprises a communication between the second person and the third person.
9. The system of claim 8 , wherein the correlation of the overlapping portions indicates the assistance of the third person in the communication between the first person and the second person.