Automated utterance search
View Patent ↗A speech analyzer is integrated or otherwise coupled to an audio player. The speech analyzer is used to identify recorded communication sessions in accordance with a search criterion. A search criterion may be spoken or otherwise communicated to the speech analyzer. Results generated by the speech analyzer are converted into visual information that is presented to a user of the speech analyzer. Results generated by the speech analyzer can be cached for real-time user review while the speech analyzer processes additional stored conversations.
1. A system for analyzing voice-based communication sessions, comprising:
a player configured to replay a recording of a communication session;
an analysis engine coupled to the player, the analysis engine configured to process a stored communication session to generate a result responsive to an utterance-of-interest in the stored communication session, wherein the analysis engine uses a language model specific to a speaker when an identity of the speaker is known;
a presentor coupled to the analysis engine and configured to present a representation of the result; and
a storage device coupled to the analysis engine, the storage device configured to cache a representation of the result, wherein the representation of the result comprises an indication of a match of the utterance-of-interest in the stored communication session and an indication of a certainty that the match is an actual match of the utterance-of-interest.
2. The system of claim 1 , wherein the representation further comprises one of a text file, a database entry, and an alternative format.
3. The system of claim 1 , wherein the analysis engine is responsive to a configuration parameter that communicates a function of result accuracy and a processing rate.
4. The system of claim 1 , wherein the presentor presents a visual representation of the location of the match in the stored communication session.
5. The system of claim 4 , wherein the model is responsive to metadata concerning the stored communication session.
6. The system of claim 1 , wherein the analysis engine is responsive to a spoken search criterion.
7. The system of claim 6 , wherein the spoken search criterion is entered by a party whose voice is represented in the stored communication session.
8. The system of claim 1 , wherein the analysis engine comprises one of a phonetic analyzer and a large vocabulary speech recognition analyzer.
9. The system of claim 1 , further comprising:
an automated performance manager coupled to the analysis engine and configured to receive information responsive to an identified agent.
10. The system of claim 9 , wherein the automated performance is configured to generate at least one agent quality score.
11. The system of claim 1 , wherein the presentor presents a visual representation of each stored communication session that contains a match with the search criterion.
12. The system of claim 11 , wherein the presentor presents a visual indicator having a characteristic that varies as a function of a confidence level in the match.
13. A method for enhancing an audio player, comprising:
integrating a speech analysis engine with the audio player;
verbally communicating a first search criterion to the speech analysis engine;
using the speech analysis engine to identify a recorded communication session in response to the first search criterion by processing a select communication session with the speech analysis engine to generate a result, wherein the speech analysis engine uses a language model specific to a speaker when an identity of the speaker is known;
using the speech analysis engine to cache a representation of the result, wherein the representation of the result comprises an indication of a match of the first search criterion in the select communication session and an indication of a certainty that the match is an actual match of the first search criterion; and
translating the result into a visual representation.
14. The method of claim 13 , wherein using the speech analysis engine receives a second search criterion via a non-voiced communication medium.
15. The method of claim 13 , wherein translating the result into a visual representation comprises generating an icon having a characteristic that varies as a function of a confidence level in the match.
16. The method of claim 13 , further comprising:
forwarding information responsive to the result and an identified agent to a performance manager.
17. A method for analyzing a communication session, comprising:
integrating a speech analysis engine with a recorder/player;
using the recorder/player to record a set of communication sessions;
identifying a subset of the set of communication sessions to analyze;
communicating a search criterion indicative of an utterance-of-interest to the speech analysis engine;
using the player portion of the recorder/player to communicate each member of the subset of communication sessions to the speech analysis engine;
using the speech analysis engine to identify a match of the utterance-of-interest in a member of the subset of communication sessions, wherein the speech analysis engine uses a language model specific to a speaker when an identity of the speaker is known; and
using the speech analysis engine to cache a result, wherein the result comprises an indication of the match of the utterance-of-interest in the member of the subset of communication sessions and an indication of a certainty that the match is an actual match of the utterance-of-interest.
18. The method of claim 17 , wherein the match of the utterance-of-interest is used as an input applied to an agent-evaluation process.
19. The method of claim 17 , wherein the match of the utterance-of-interest is used as an input applied to a fraud-detection process.
20. The method of claim 19 , wherein the fraud detection process generates an alert.
21. The method of claim 19 , wherein the fraud detection process identifies a speaker.
22. The method of claim 17 , wherein the match is used as an input applied to an e-learning tool.