https://www.cs.swarthmore.edu/~turnbull/cs97/f09/paper/reynolds00.pdf

Speaker Verification Using Adapted Gaussian Mixture Models

The performance of the MUSIC and ML methods is studied, and their statistical efficiency is analyzed. The Cramer-Rao bound (CRB) for the estimation problems is derived, and some useful properties of the CRB covariance matrix are established. The relationship between the MUSIC and ML estimators is investigated as well. A numerical study is reported of the statistical efficiency of the MUSIC estimator for the problem of finding the directions of two plane waves using a uniform linear array. An exact description of the results is included. >

MUSIC, maximum likelihood, and Cramer-Rao bound

A study of a matrix pencil method for estimating frequencies and damping factors of exponentially damped and/or undamped sinusoids in noise is presented. Comparison of this method to a polynomial method (SVD-Prony method) shows that the matrix pencil method and the polynomial method are two special cases of a matrix prediction approach and that the pencil method is more efficient in computation and less restrictive about signal probes. It is found through perturbation analysis and simulation that, for signals with unknown damping factors, the pencil method is less sensitive to noise than the polynomial method. An expression of the Cramer-Rao bound for the exponential signals is presented. >

/pdf/matrix-pencil-method-for-estimating-parameters-of-s9z3xz6605.pdf

Matrix pencil method for estimating parameters of exponentially damped/undamped sinusoids in noise

An intelligent automated assistant system engages with the user in an integrated, conversational manner using natural language dialog, and invokes external services when appropriate to obtain information or perform various actions. The system can be implemented using any of a number of different platforms, such as the web, email, smartphone, and the like, or any combination thereof. In one embodiment, the system is based on sets of interrelated domains and tasks, and employs additional functionally powered by external services with which the system can interact.

Intelligent Automated Assistant

Systems and methods are provided for receiving speech and non-speech communications of natural language questions and/or commands, transcribing the speech and non-speech communications to textual messages, and executing the questions and/or commands. The invention applies context, prior information, domain knowledge, and user specific profile data to achieve a natural environment for one or more users presenting questions or commands across multiple domains. The systems and methods creates, stores and uses extensive personal profile information for each user, thereby improving the reliability of determining the context of the speech and non-speech communications and presenting the expected results for a particular question or command.

Systems and methods for responding to natural language speech utterance

Speech recognition models are dynamically re-configurable based on user information, application information, background information such as background noise and transducer information such as transducer response characteristics to provide users with alternate input modes to keyboard text entry. Word recognition lattices are generated for each data field of an application and dynamically concatenated into a single word recognition lattice. A language model is applied to the concatenated word recognition lattice to determine the relationships between the word recognition lattices and repeated until the generated word recognition lattices are acceptable or differ from a predetermined value only by a threshold amount. These techniques of dynamic re-configurable speech recognition provide for deployment of speech recognition on small devices such as mobile phones and personal digital assistants as well environments such as office, home or vehicle while maintaining the accuracy of the speech recognition.

System and method of performing user-specific automatic speech recognition

Likelihood ratio or cohort normalized scoring has been shown to be effective for improving the performance of speaker verification systems. An important problem in this connection is the establishment of principles for constructing speaker background or cohort models which provide the most effective normalized scores. Several kinds of speaker background models are studied. These include individual speaker models, models constructed from the pooled utterances of different numbers of speakers, models selected on the basis of similarity with customer models, models constructed from random selections of speakers, and models constructed from databases recorded under different conditions than the customer models. The results of experiments show that pooled models based on similarity to the reference speaker perform better than individual cohort models from the same similar set of speakers. Pooled background models from a small number of speakers based on similarity perform about the best, but not significantly better than a random selection of 40 or more gender balanced speakers with training conditions matched to the reference speakers.

Speaker background models for connected digit password speaker verification

Recognition of spoken names is a challenging task for speech recognition systems because of the large variations in speaking styles, linguistic origins and pronunciation found in names. The complex linguistic nature of names makes it difficult to automatically generate pronunciation variations. For many applications the list of names tends to be in the order of several hundred thousands, making spoken name recognition a high perplexity task. Use of multiple pronunciations to account for the variations in names further increases the perplexity of the recognition system substantially. In this paper we propose the use of the syllable as the acoustic unit for spoken name recognition and show how pronunciation variation modeling with syllables can help in improving recognition performance and reducing the system perplexity. We present results comparing systems which use context dependent phones with syllable based systems, and demonstrate that a significant increase in recognition accuracy and speed, can be achieved by using the syllable as the acoustic unit for spoken name recognition. With a finite state grammar network for spoken name recognition, the observed recognition error rate for the syllable-based system was 40% less than the phone-based system. For syllable bigram based information retrieval schemes the observed recognition error rate was about 60% less than the corresponding phone system.

A syllable-based approach for improved recognition of spoken names

This paper describes the AT&T WATSON real-time speech recognizer, the product of several decades of research at AT&T. The recognizer handles a wide range of vocabulary sizes and is based on continuous-density hidden Markov models for acoustic modeling and finite state networks for language modeling. The recognition network is optimized for efficient search. We identify the algorithms used for high-accuracy, real-time and low-latency recognition. We present results for small and large vocabulary tasks taken from the AT&T VoiceTone/sup /spl reg// service, showing word accuracy improvement of about 5% absolute and real-time processing speed-up by a factor between 2 and 3.

/pdf/the-at-t-watson-speech-recognizer-28ez551pdl.pdf

The AT&T WATSON speech recognizer

A new pronunciation learning system for dynamically learning new pronunciations assisted by user correction logs. The user correction logs provide a record of speech recognition events and subsequent user behavior that implicitly confirms or rejects the recognition result and/or shows the user's intended words by via subsequent input. The system analyzes the correction logs and distills them down to a set of words which lack acceptable pronunciations. Hypothetical pronunciations, constrained by spelling and other linguistic knowledge, are generated for each of the words. Offline recognition determines the hypothetical pronunciations with a good acoustical match to the audio data likely to contain the words. The matching pronunciations are aggregated and adjudicated to select new pronunciations for the words to improve general or personalized recognition models.

Sarangarajan Parthasarathy

Papers

System and method of performing user-specific automatic speech recognition

Speaker background models for connected digit password speaker verification

A syllable-based approach for improved recognition of spoken names

The AT&T WATSON speech recognizer

Pronunciation learning through correction logs