Abstract: A unified clustering tree (500) generates phoneme clusters based on an input sequence of phonemes. The number of possible clusters is significantly less than the number of possible combinations of input phonemes. Nodes (510, 511) in the unified clustering tree are arranged into levels such that the clustering tree generates clusters for multiple speech recognition models. Models that correspond to higher levels in the unified clustering tree are coarse models relative to more fine-grain models at lower levels of the clustering tree.

Abstract: A personal information manager (PIM) [100] stores user personal data. The PIM may be a personal computer like box located in the home of the user. The PIM includes an audio interface [108] and a visual interface [110]. Users may equivalently interact and manage their data through the audio interface or the visual interface. Accordingly, the user has full access to the PIM whenever the user can establish a voice or data connection.

Abstract: A linguistic segmentation tool (115) includes an acoustic feature extraction component (302) and a lexical feature extraction component (311). The acoustic feature extraction component (302) extracts prosodic features from speech (e.g., pauses, pitch, energy, and rate). The lexical feature extraction component (311) extracts lexical features from a transcribed version of the speech (e.g., words, syntactic classifications of the words, and word structure). A language model is constructed based on the lexical features and an acoustic model is constructed based on the acoustic features. A statistical framework combines the outputs of the language model to generate indications of potential linguistic features.

Abstract: A thematic segmentation tool generates indications of thematically coherent segments within a document. The thematic segmentation tool includes a transcription component, a speaker boundary detection component, a linguistic detection component, and a topic classification component. Each of these components generates input information for a thematic decision component, which generates the thematic segmentation information. Multiple thematic segments may concurrently apply to a portion of a document. Additionally, the thematic segments may be hierarchical thematic segments.

Abstract: An automated meeting facilitator [109] manages and archives a telemeeting. The automated meeting facilitator includes a multimedia indexing section [220], a memory section [230], and a server [240]. The automated meeting facilitator may connect to meeting participants [102] through a network. The multimedia indexing section [220] generates rich transcriptions of the telemeeting and stores documents related to the telemeeting. Through the rich transcription, the automated meeting facilitator is able to provide a number of real-time search and assistance functions to the meeting participants.

Abstract: A system aids a user in translating an audio signal that includes speech from one language to another. The system retrieves a textual representation of the audio signal and presents the textual representation to the user. The system receives selection of a segment of the textual representation for translation and obtains a portion of the audio signal corresponding to the segment. The system then provides the segment of the textual representation and the portion of the audio signal to the user to help the user translate the audio signal.

Abstract: A transcription tool assists a user in transcribing audio. The transcription tool includes an audio classification component that classifies an incoming audio stream based on whether portions of the audio stream contain speech data. The transcription tool plays the portions of the audio that contain speech data back to the user and skips the portions of the audio that do not contain speech data. Using a relatively simple command set, the user can control the transcription tool and annotate transcribed text.