Adapting to adverse acoustic environment in speech processing using playback training data

#This patent document contains information subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent, as it appears in the U.S. Patent and Trademark Office files or records but otherwise reserves all copyright rights whatsoever.

#Aspects of the present invention relate to automated speech processing. Other aspects of the present invention relate to adaptive automatic speech recognition.

#In a society that is becoming increasingly “information anywhere and anytime”, voice enabled solutions are often deployed to provide voice information services. For example, a telephone service company may offer a voice enabled call center so that inquiries from customers may be automatically directed to appropriate agents. In addition, voice information services may be necessary to users who communicate using devices that do not have a platform on which information can be exchanged in conventional textual form. In these applications, an automatic speech recognition system may be deployed to enable voice-based communications between a user and a service provider.

#The present invention is further described in terms of exemplary embodiments, which will be described in detail with reference to the drawings. These embodiments are nonlimiting exemplary embodiments, in which like reference numerals represent similar parts throughout the several views of the drawings, and wherein:

#FIG. 1 depicts a framework that facilitates an automated speech recognition mechanism to adapt to an adverse acoustic environment based on playback speech training data, according to embodiments of the present invention;

#FIG. 2 illustrates exemplary types of automatic speech recognition models;

#The processing described below may be performed by a properly programmed general-purpose computer alone or in connection with a special purpose computer. Such processing may be performed by a single platform or by a distributed processing platform. In addition, such processing and functionality can be implemented in the form of special purpose hardware or in the form of software being run by a general-purpose computer. Any data handled in such processing or created as a result of such processing can be stored in any memory as is conventional in the art. By way of example, such data may be stored in a temporary memory, such as in the RAM of a given computer system or subsystem. In addition, or in the alternative, such data may be stored in longer-term storage devices, for example, magnetic disks, rewritable optical disks, and so on. For purposes of the disclosure herein, a computer-readable media may comprise any form of data storage mechanism, including such existing memory technologies as well as hardware or circuit representations of such structures and of such data.

#FIG. 1 depicts a framework 100 that facilitates an automated speech recognition mechanism 180 to adapt to an adverse acoustic environment, according to embodiments of the present invention. The framework 100 comprises an automatic speech recognition mechanism 105, a training data sampling and playback mechanism 140, a re-recording mechanism 150, and an adverse acoustic environment adaptation mechanism 160. The automatic speech recognition mechanism 105 takes input speech 102 as input and generates output text 104 as output. The output text 104 represents the textual information automatically transcribed, by the automatic speech recognition mechanism 105, from the input speech 102.

#The automatic speech recognition mechanism 105 includes a set of training data 110, an automatic speech recognition (ASR) model generation mechanism 120, a set of ASR models 130, an automatic speech recognizer 180, and an adaptation mechanism 170. Automatic speech recognition is usually performed according to some pre-established models (e.g., ASR models 130). Such models may characterize speech with respect to different aspects. For example, the linguistic aspect of speech may be modeled so that the grammatical structure of a language may be properly characterized. The pronunciation aspect of speech may be characterized using an acoustic model, which is specific to a particular language.