System and method for three-way call detection
A system for detecting three-way calls in a monitored telephone conversation includes a speech recognition processor that transcribes the monitored telephone conversation and associates characteristics of the monitored telephone conversation with a transcript thereof, a database to store the transcript and the characteristics associated therewith, and a three-way Call detection processor to analyze the characteristics of the conversation and to detect therefrom the addition of one or more parties to the conversation. The system preferably includes at least one domain-specific language model that the speech recognition processor utilizes to transcribe the conversation. The system may operate in real-time or on previously recorded conversations. A query and retrieval system may be used to retrieve and review call records from the database.
1. A system for detecting a three-way call in a monitored telephone conversation, the system comprising:
one or more circuits configured to:
generate a domain specific language model associated with a participant of the monitored telephone conversation;
convert an audio of the monitored telephone conversation to a transcript using the domain specific language model;
detect an utterance of a phrase occurring at a first time within the monitored telephone conversation based on the transcript of the monitored telephone conversation;
determine an audio characteristic occurring at a second time within the monitored telephone conversation;
detect a pattern based on the phrase, the audio characteristic, and a comparison of the first time and the second time; and
detect the three-way call based on the pattern, wherein the pattern is an indicator of the three-way call.
2. The system of claim 1 , wherein the one or more circuits are further configured to: tag the transcript with a timestamp associated with detection of the three-way call in the monitored telephone conversation.
3. The system of claim 1 , wherein the one or more circuits are further configured to: store at least one of the audio and the transcript to a database.
4. The system of claim 1 , wherein to convert the audio of the monitored telephone conversation to the transcript using the domain specific language model, the one or more circuits are configured to:
translate the transcript of the monitored telephone conversation from a first language into a second language.
5. The system of claim 1 , wherein to detect the three-way call based on the pattern, the one or more circuits are configured to:
determine a value corresponding to the pattern; and
determine that the value is greater than a predetermined threshold.
6. The system of claim 1 , wherein to generate the domain specific language model associated with the participant of the monitored telephone conversation, the one or more circuits are configured to train at least one of:
the domain specific language model to transcribe inmate telephony;
the domain specific language model to transcribe a foreign language; or
the domain specific language model to transcribe a linguistic dialect.
7. The system of claim 1 , wherein the pattern includes one or more of a frequency of the utterance of the phrase during the monitored telephone conversation, a timing of the utterance of the phrase during the monitored telephone conversation, or an entity extracted from the utterance of the phrase.
8. The system of claim 1 , wherein the monitored telephone conversation is one of a monitored voice over IP (VOIP) telephone call or a monitored plain old telephone system (POTS).
9. A method for detecting a three-way call in a monitored telephone conversation, the method comprising:
generating a domain specific language model associated with a participant of the monitored telephone conversation;
converting an audio of the monitored telephone conversation to a transcript using the domain specific language model;
detecting an utterance of a phrase occurring at a first time within the monitored telephone conversation based on the transcript of the monitored telephone conversation;
determining an audio characteristic occurring at a second time within the monitored telephone conversation;
detecting a pattern based on the phrase, the audio characteristic, and a comparison of the first time and the second time; and
detecting the three-way call based on the pattern, wherein the pattern is an indicator of the three-way call.
10. The method according to claim 9 , further comprising: storing at least one of the audio and the transcript to a database.
11. The method according to claim 9 , further comprising:
tagging the transcript with a timestamp associated with detection of the three-way call in the monitored telephone conversation.
12. The method according to claim 11 , further comprising:
providing a synchronized presentation of the audio and the transcript.
13. The method according to claim 9 , wherein converting the spoken audio of the monitored telephone conversation to the transcript using the domain specific language model comprises:
translating the transcript of the monitored telephone conversation from a first language into a second language.
14. The method according to claim 9 , wherein detecting the three-way call based on the pattern comprises:
determining a value corresponding to the pattern; and
determining that the value is greater than a predetermined threshold.
15. The method according to claim 9 , wherein generating the domain specific language model associated with the participant of the monitored telephone conversation comprises at least one of:
training the domain specific language model to transcribe inmate telephony;
training the domain specific language model to transcribe a foreign language; or
training the domain specific language model to transcribe a linguistic dialect.
16. The method according to claim 9 , wherein the pattern includes one or more of a frequency of the utterance of the phrase during the monitored telephone conversation, a timing of the utterance of the phrase during the monitored telephone conversation, or an entity extracted from the utterance of the phrase.
17. A non-transitory computer-readable device having instructions stored thereon that, when executed by at least one computing device, causes the at least one computing device to perform operations comprising:
generating a domain specific language model associated with a participant of a monitored telephone conversation;
converting an audio of the monitored telephone conversation to a transcript using the domain specific language model;
detecting an utterance of a phrase occurring at a first time within the monitored telephone conversation based on the transcript of the monitored telephone conversation;
determining an audio characteristic occurring at a second time within the monitored telephone conversation;
detecting a pattern based on the phrase, the audio characteristic, and a comparison of the first time and the second time; and
detecting the three-way call based on the pattern, wherein the pattern is an indicator of the three-way call.
18. The non-transitory computer-readable device of claim 17 , the operations further comprising:
tagging the transcript with a timestamp associated with detection of the three-way call in the monitored telephone conversation.
19. The non-transitory computer-readable device of claim 18 , the operations further comprising:
providing a synchronized presentation of the audio and the transcript.
20. The non-transitory computer-readable device of claim 17 , wherein the pattern includes one or more of a frequency of the utterance of the phrase during the monitored telephone conversation, a timing of the utterance of the phrase during the monitored telephone conversation, or an entity extracted from the utterance of the phrase.