International Business Machines Corporation

Interactive Intelligence Group, Inc.

Conduent Business Services, LLC

Other

Xerox Corporation

Interactive Intelligence, Inc.

Vivek Tyagi

Felix Immanuel Wyss

Aravind Ganapathiraju

Biplav Srivastava

Raghuram Krishnapuram

Tanveer Afzal Faruquie

Ananth Nagaraja Iyer

Prathosh Aragulla Prasad

Scott Allen Randal

Zhenhao Ge

Nandita J Mahajan

L Venkat Subramaniam

Shivkumar Kalyanaraman

Theja Tulabandhula

Ullas Balan Nambiar

Smruti Ranjan Sarangi

Niranjan Aniruddha Viladkar

A system and method are presented for acoustic data selection of a particular quality for training the parameters of an acoustic model, such as a Hidden Markov Model and Gaussian Mixture Model, for example, in automatic speech recognition systems in the speech analytics field. A raw acoustic model may be trained using a given speech corpus and maximum likelihood criteria. A series of operations are performed, such as a forced Viterbi-alignment, calculations of likelihood scores, and phoneme recognition, for example, to form a subset corpus of training data. During the process, audio files of a quality that does not meet a criterion, such as poor quality audio files, may be automatically rejected from the corpus. The subset may then be used to train a new acoustic model.

Method and system for acoustic data selection for training the parameters of an acoustic model

What is disclosed is a system and method for modelling a class posterior probability of context dependent phonemes in a speech recognition system. A representation network is trained by projecting a N-dimensional feature vector into G intermediate layers of nodes. At least some features are associated with a class label vector. A last intermediate layer Zof the representation network is discretized to obtain a discretized layer {circumflex over (Z)}. Feature vector Q is obtained by randomly selecting V features from discretized layer {circumflex over (Z)}. Q is repeatedly hashed to obtain a vector Qwhere Qis an output of the fhashing. An equivalent scalar representation is determined for each Q. In a manner more fully disclosed herein, a posterior probability Pis determined for each (x, b) pair based on the equivalent scalar representation of each respective Q. The obtained posterior probabilities are used to improve classification accuracy in a speech recognition system.

Modeling a class posterior probability of context dependent phonemes in a speech recognition system

A method and a system for detecting sentiment of a human based on an analysis of human speech are disclosed. In an embodiment, one or more time instances of glottal closure are determined from a speech signal of the human. A voice source signal based on the determined one or more time instances of glottal closure is generated. A set of relative harmonic strengths is determined based on one or more harmonic contours of the voice source signal. The RHS is indicative of a deviation of the one or more harmonics of the voice source signal from a fundamental frequency of the voice source signal. A set of feature vectors is determined based on the RHS. The set of feature vectors are utilizable to detect the sentiment of the human.

Method and system for detecting sentiment by analyzing human speech

The disclosed embodiments relate to a method of keyword recognition in a speech signal. The method includes determining a first likelihood score and a second likelihood score of one or more features of a frame of said speech signal being associated with one or more states in a first model and one or more states in a second model, respectively. The one or more states in the first model corresponds to one or more tied triphone states and the one or more states in the second model corresponds to one or more monophone states of a keyword to be recognized in the speech signal. The method further includes determining a third likelihood score based on the first likelihood score and the second likelihood score. The first likelihood score and the third likelihood score are utilizable to determine presence of the keyword in the speech signal.

Methods and systems for identifying keywords in speech signal

A system and method for learning alternate pronunciations for speech recognition is disclosed. Alternative name pronunciations may be covered, through pronunciation learning, that have not been previously covered in a general pronunciation dictionary. In an embodiment, the detection of phone-level and syllable-level mispronunciations in words and sentences may be based on acoustic models trained by Hidden Markov Models. Mispronunciations may be detected by comparing the likelihood of the potential state of the targeting pronunciation unit with a pre-determined threshold through a series of tests. It is also within the scope of an embodiment to detect accents.

System and method for learning alternate pronunciations for speech recognition

A system and method are presented for selectively biased linear discriminant analysis in automatic speech recognition systems. Linear Discriminant Analysis (LDA) may be used to improve the discrimination between the hidden Markov model (HMM) tied-states in the acoustic feature space. The between-class and within-class covariance matrices may be biased based on the observed recognition errors of the tied-states, such as shared HMM states of the context dependent tri-phone acoustic model. The recognition errors may be obtained from a trained maximum-likelihood acoustic model utilizing the tied-states which may then be used as classes in the analysis.

Method and system for selectively biased linear discriminant analysis in automatic speech recognition systems

Methods and arrangements for accommodating and synthesizing data for seeking broad-based solutions to complex problems. Logical models with respect to infrastructure in a first location are created. System-level models are created with respect to the logical models, and simulations are performed on the system-level models to verify expected behavior. Feedback is employed from the performed simulations to forecast changes in the system-level models.

Systems and methods for exploring and utilizing solutions to cyber-physical issues in a sandbox

Methods and arrangements for employing roadside acoustics sensing in ascertaining traffic density states. Traffic monitoring input is received from a road segment, the traffic monitoring input including traffic audio input. The traffic monitoring input is processed and the processed traffic monitoring input is classified with a predetermined traffic density state. The classified traffic monitoring input is combined with other classified traffic monitoring input.

Systems and methods for road acoustics and road video-feed based traffic estimation and prediction

Techniques for protecting information in an audio file are provided. The techniques include obtaining an audio file, detecting information beating one or more segments in a speech signal, wherein the information comprises information sought for protection, encrypting the information sought for protection by scrambling the one or more segments using a scrambling filter, and selectively decrypting an amount of the encrypted information, wherein the amount of the encrypted information to be decrypted depends on user access privilege, and wherein selectively decrypting the amount of the encrypted information protects said amount of the encrypted information. Techniques are also provided for protecting information in an audio file.

System and computer program product for protecting audio content

Techniques for protecting information in an audio file are provided. The techniques include obtaining an audio file, detecting information bearing one or more segments in a speech signal, wherein the information comprises information sought for protection, encrypting the information sought for protection by scrambling the one or more segments using a scrambling filter, and selectively decrypting an amount of the encrypted information, wherein the amount of the encrypted information to be decrypted depends on user access privilege, and wherein selectively decrypting the amount of the encrypted information protects said amount of the encrypted information. Techniques are also provided for protecting information in an audio file.

Growing community of inventors

New Delhi, India

Vivek Tyagi