Abstract: Systems and methods for assisting automatic speech recognition (ASR) are provided. An example method includes generating, by a mobile device, a plurality of instantiations of a speech component in a captured audio signal, each instantiation of the plurality of instantiations being in support of a particular hypothesis regarding the speech component. At least two instantiations of the plurality of instantiations are then sent to a remote ASR engine. The remote ASR engine is configured to recognize at least one word based on the at least two of the plurality of instantiations and a user context, according to various embodiments. This recognition can include selecting one of the instantiations of the speech component from the plurality of instantiations. The plurality of instantiations may be generated by noise suppression of the captured audio signal with different degrees of aggressiveness.

Abstract: Systems and methods for registration of a customer inside a business place are disclosed. A method can include transmitting, by one or more ultrasound speakers associated with the business place, an ultrasound audio signature to a mobile device associated with a customer. In addition, the method can include sending, over one or more networks, a message from the mobile device to a server. An example message includes at least information associated with the customer and the ultrasound audio signature. In response to the message, a notification concerning the customer is sent from the server to the business place, according to various embodiments.

Abstract: A method for performing time and frequency Signal-to-Noise Ratio (SNR) dependent weighting in speech recognition is described that includes for each period t estimating the SNR to get time and frequency SNR information ?t,f; calculating the time and frequency weighting to get ?tf; performing the back and forth weighted time varying DCT transformation matrix computation MGtM?1 to get Tt; providing the transformation matrix computation Tt and the original MFCC feature ot that contains the information about the SNR to a recognizer including the Viterbi decoding; and performing weighted Viterbi recognition bj(ot).

Abstract: The operating range of joint additive and convolutive compensating method is extended by enhanced channel estimation procedure that adds SNR-dependent inertia and SNR-dependent limit on the channel estimate.

Abstract: A method for training acoustic models for a new target language is provided using a phonetic table, which characterizes the phones, used in one or more reference language(s) with respect to their articulatory properties; a phonetic table, which characterizes the phones used in the target language with respect to their articulatory properties; a set of trained monophones for the reference language(s); and a database of sentences in the target language and its phonetic transcription. With these inputs, the new method completely and automatically takes care of the steps of monophone seeding and triphone clustering and machine intensive training steps involved in triphone acoustic training.

Abstract: A method and detector for providing a noise resistant utterance detector is provided by extracting a noise estimate (15) to augment the signal-to-noise ratio of the speech signal, inverse filtering (17) of the speech signal to focus on the periodic excitation part of the signal and spectral reshaping (19) to accentuate separation between formants.