Abstract: A method, system, and computer program product for spoken language grammar evaluation are provided. The method includes playing a recorded question to a candidate, recording a spoken answer from the candidate, and converting the spoken answer into text. The method further includes comparing the text to a grammar database, calculating a spoken language grammar evaluation score based on the comparison, and outputting the spoken language grammar evaluation score.

Abstract: A method, system, and computer program product for spoken language grammar evaluation are provided. The method includes playing a recorded question to a candidate, recording a spoken answer from the candidate, and converting the spoken answer into text. The method further includes comparing the text to a grammar database, calculating a spoken language grammar evaluation score based on the comparison, and outputting the spoken language grammar evaluation score.

Abstract: A phonetic vocabulary for a speech recognition system is adapted to a particular speaker's pronunciation. A speaker can be attributed specific pronunciation styles, which can be identified from specific pronunciation examples. Consequently, a phonetic vocabulary can be reduced in size, which can improve recognition accuracy and recognition speed.

Abstract: A phonetic vocabulary for a speech recognition system is adapted to a particular speaker's pronunciation. A speaker can be attributed specific pronunciation styles, which can be identified from specific pronunciation examples. Consequently, a phonetic vocabulary can be reduced in size, which can improve recognition accuracy and recognition speed.

Abstract: The present invention provides a hybrid call handling method and system. The method comprises navigating a plurality of received calls from a plurality of callers. The method further comprises monitoring a call health status for each of the plurality of the calls being navigated for entire call duration and notifying a bad call health status of the monitored call to a human agent for employing at least one rectification action. The call health status is determined by monitoring and measuring one or more call parameters. The invention provides for a system for call handling and navigation by an automated system with a human agent assisting the automated system for rectification of calls with bad call health status. Once the call with a bad health is transferred to the human agent, he assists the automated system either by directly communicating with the caller or by communicating using a machine interface.

Abstract: A mobile wireless device, such as a mobile wireless phone, is adapted based on a user's current abstracted contextual situation, where the context of a user is determined using devices enabled with near-field communication technology. Dynamic information of a user of a mobile device, such as the identity of his or her current environment, is determined using near-field communication, such as radio-frequency identification (RFID) tags. Static information regarding the user is also determined, where such static information can include the user's preferences regarding how the mobile device should adapt to certain environments. An abstracted contextual situation of the user is synthesized based on this dynamic and static information. One or more adaptation directives for the mobile device of the user are determined based on the user's abstracted contextual situation.

Abstract: Systems and methods for stripping an optical mode from a semiconductor laser. A waveguide layer with multiple layers is included in the semiconductor laser and is typically arranged beneath the active region. The waveguide layer is configured to match the phase of the second order mode. The waveguide layer does not substantially match the primary optical mode of the laser. By matching the phase of the second order mode, the confinement of the second order mode is reduced and the second order mode strongly couples with the waveguide layer. The optical confinement of the primary mode is not substantially reduced. The side-mode suppression ratio is thereby improved by stripping the second order mode from the active region.

Abstract: Bootstrapping of a system from one language to another often works well when the two languages share the similar acoustic space. However, when the new language has sounds that do not occur in the language from which the bootstrapping is to be done, bootstrapping does not produce good initial models and the new language data is not properly aligned to these models. The present invention provides techniques to generate context dependent labeling of the new language data using the recognition system of another language. Then, this labeled data is used to generate models for the new language phones.

Abstract: A method, a computer system and a computer program product for generating baseforms or phonetic spellings from input text are disclosed. The baseforms are initially generated using rules defined for a particular language. Then, phones are identified in the language that are exceptions to the defined rules and an action is associated with each identified phone. A statistical technique is applied to determine whether the identified phones can be modified. Finally, baseforms containing the identified phones that can be modified, are corrected according to the associated actions. Preferably, the statistical technique is only applied to baseforms containing phones that are exceptions to the defined rules. The defined rules can comprise spelling-to-sound rules for a particular phonetic language that incorporate all possible alternative pronunciations of each baseform.

Abstract: Speech utterance waveforms are visually compared in which syllables are indicated, such as in color. A speech utterance from a first user and a corresponding speech utterance from a second user are recorded. The phones, or phonemes, of each speech utterance are segmented, and these phones are mapped to the syllables of the speech utterances. A waveform of each speech utterance is displayed, in which syllables of the words spoken in the speech utterance are indicated. The syllables of the words may be distinguished in different colors, such that the same color is used for the same syllable in both of the utterances. A specific color can also be used to specify the stress level of a syllable. The users may visually compare the waveforms to assist understanding of the differences in syllable stress patterns between the utterance of the first user and the corresponding utterance of the second user.

Abstract: Systems and methods for short length gain regions in edge emitting lasers. After forming the lower layers of a laser on a substrate, the active layer is formed. The active layer is then selectively etched. The unetched portion of the active layer corresponds to an active region of the laser. The etched portions are then selectively regrown with a material that is transparent to light emitted by the active region. The active layer thus includes an active region and an inactive region. Next, the upper layers are grown or formed over the active layer. Selective regrowth of the active layer enables a length of the active region to be independent of the cleaved length of the laser. This reduces current and power requirements of the laser.

Abstract: Methods and apparatuses for wafer testing edge emitting lasers and providing a vertical emission from an edge emitting laser. A plurality of edge emitting lasers can be formed on a semiconductor wafer. One or more grooves can be etched into the semiconductor wafer to form etched facets for the edge emitting lasers. A current can be applied to at least one edge emitting laser to produce at least one optical output. An evaluation of the at least one optical output can be performed while the edge emitting lasers are still in wafer form. Edge emitting lasers may also be produced including a reflective surface for reflecting at least one edge emitted optical signal in a vertical perpendicular direction. The reflective surface can be created using an etching process during manufacture of the edge emitting laser.

Abstract: This disclosure concerns a vertical lasing semiconductor optical amplifier (VLSOA) having a quantum dot active region. In one example, a VLSOA includes a quantum dot active region comprising a semiconductor gain medium. The semiconductor gain medium defines at least a portion of an amplifying path. The VLSOA also includes a laser cavity within which a portion of the semiconductor gain medium is disposed. The laser cavity has a gain characteristic, with respect to an optical signal traversing the amplifying path, that is responsive to a pump input to the laser cavity.

Abstract: Systems and methods for controlling edge gain in avalanche photodiodes. During fabrication of an avalanche photodiode, the photodiode is diffused with a dopant. The mask used for the dopant includes a plurality of openings such that the dopant diffuses within the photodiode to create a plurality of interconnected spheres. The diffusion front has a shape to introduce an edge effect into the center of the photodiode. The diffusion front ameliorates the edge effect by introducing the edge effect into the center of the photodiode.

Abstract: Systems and methods for tuning a DBR stack for an optical amplifier. The DBR layers in a mirror of the optical amplifier have a duty cycle that can be altered to tune a location of a channel drop in a gain spectrum. In addition to changing the duty cycle, the DBR stacks can be segmented. The segmented DBR stacks and/or the selected duty cycle tunes a location of a channel drop outside of a range of wavelengths of interest.

Abstract: Systems and methods for extending a linear range of a semiconductor optical amplifier (SOA). A feedback layer is included in an SOA. The optical mode of the SOA is distributed between the feedback layer and the active region. As output optical power increases, the mode confinement of the active region increases and the mode is drawn from the feedback layer into the active region. The increase in the mode confinement offsets a loss of material gain such that the linear range of the SOA is extended. In one embodiment, the modal gain increases an higher output optical powers.

Abstract: Systems and methods for stripping an optical mode from a semiconductor laser. A waveguide layer is included in the semiconductor laser and is typically arranged beneath the active region. The waveguide layer is configured to match the phase of the second order mode. The waveguide layer does not substantially match the primary optical mode of the laser. By matching the phase of the second order mode, the confinement of the second order mode is reduced and the second order mode strongly couples with the waveguide layer. The optical confinement of the primary mode is not substantially reduced. The side-mode suppression ratio is thereby improved by stripping the second order mode from the active region.

Abstract: A method of classifying samples to one of a number of predetermined classes involves using a number of class models or classifiers to form order statistic for each classifier. A linear combination of the order statistic (L-statistic) is calculated to determine the confidence of that particular classifier, both in general and for that particular sample. Relative weights are then derived from these confidences, and used to calculate a weighted summation across all classifiers for each class of the likelihoods that a sample belongs to that class. The sample is classified in the class which has the associated weighted summation which is greatest in value.

Abstract: A language model is constructed for mixed language expressions that have words from more than one natural language. Word equivalence probabilities for pairs of words among the languages are generated and stored. Word equivalence probabilities are used as required to generate a monolingual word history. The monolingual history is used by a monolingual language model to generate a next-word hypothesis. The word equivalence probabilities are also used to compute the next word probabilities in the foreign language.

Abstract: A computer implemented method in a language independent system generates audio-driven facial animation given the speech recognition system for just one language. The method is based on the recognition that once alignment is generated, the mapping and the animation hardly have any language dependency in them. Translingual visual speech synthesis can be achieved if the first step of alignment generation can be made speech independent. Given a speech recognition system for a base language, the method synthesizes video with speech of any novel language as the input.