Method and apparatus for the automatic separating and indexing of multi-speaker conversations
First Claim
1. A method of processing a continuous audio stream containing human speech related to at least one particular transaction, comprising the steps of:
- digitizing the continuous audio stream;
detecting a speaker change in the digitized audio stream;
performing a speaker recognition if a speaker change is detected;
transcribing at least part of the continuous audio stream if a predetermined speaker is recognized.
2 Assignments
0 Petitions
Accused Products
Abstract
Disclosed are a method and apparatus for processing a continuous audio stream containing human speech in order to locate a particular speech-based transaction in the audio stream, applying both known speaker recognition and speech recognition techniques. Hereby it is enabled that only the utterances of a particular predetermined speaker are transcribed thus providing an index and a summary of the underlying dialogue(s).
In a first scenario, an incoming audio stream, e.g. a speech call from outside, is scanned in order to detect audio segments of the predetermined speaker. These audio segments are then indexed and only the indexed segments are transcribed into spoken or written language. Thus an already occurred specific transaction can be found on an endless storage media like a magnetic tape. The proposed mechanism thus makes the task of locating an audio log of a specific transaction a much more less effort.
In a second scenario, two or more speakers located in one room are using a multi-user speech recognition system (SRS). For each user there exists, a different speaker model and optionally a different dictionary or vocabulary of words already known or trained by the speech or voice recognition system. In such an environment, the invention allows to switch between different dictionaries when a first user has stopped utterance and a second user is going to start his utterance.
32 Citations
18 Claims
-
1. A method of processing a continuous audio stream containing human speech related to at least one particular transaction, comprising the steps of:
-
digitizing the continuous audio stream;
detecting a speaker change in the digitized audio stream;
performing a speaker recognition if a speaker change is detected;
transcribing at least part of the continuous audio stream if a predetermined speaker is recognized. - View Dependent Claims (3, 4, 5, 6, 7, 12, 17, 18)
-
-
2. A method of processing a continuous audio stream containing human speech related to at least one particular transaction, comprising the steps of:
-
digitizing the continuous audio stream;
detecting a speaker change in the digitized audio stream;
performing a speaker recognition if a speaker change is detected;
indexing the audio stream with respect to the detected speaker change if a predetermined speaker is recognized.
-
-
8. Apparatus for processing a continuous audio stream containing human speech related to at least one particular transaction, comprising:
-
means for predetermining at least one speaker;
means for detecting speaker changes in the audio stream;
means for recognizing the predetermined speaker in the audio stream;
means for initiating transcription of at least part of the audio stream in case of a detected speaker change and a recognized predetermined speaker. - View Dependent Claims (10, 11, 13, 14, 15)
-
-
9. Apparatus for processing a continuous audio stream containing human speech related to at least one particular transaction, comprising:
-
means for predetermining at least one speaker;
means for detecting speaker changes in the audio stream;
means for recognizing the predetermined speaker in the audio stream;
means for indexing the audio stream dependent on a detected speaker change and a recognized predetermined speaker.
-
-
16. Speech recognition or voice control system processing an incoming audio stream and having at least two speaker models and/or speaker-specific dictionaries, comprising means for detecting a speaker change in the incoming audio stream;
-
means for gathering speaker-specific information and for comparing the gathered speaker-specific information with corresponding speaker-specific information of at least one predetermined speaker thus recognizing the at least one predetermined speaker;
means for interchanging between the at least two speaker-specific dictionaries dependent on the detected speaker change and the corresponding recognized speaker.
-
Specification