Abstract: A method and system for obtaining a pool of speech syllable models. The model pool is generated by first detecting a training segment using unsupervised speech segmentation or speech unit spotting. If the model pool is empty, a first speech syllable model is trained and added to the model pool. If the model pool is not empty, an existing model is determined from the model pool that best matches the training segment. Then the existing module is scored for the training segment. If the score is less than a predefined threshold, a new model for the training segment is created and added to the pool. If the score equals the threshold or is larger than the threshold, the training segment is used to improve or to re-estimate the model.

Abstract: A password grammar for speech recognition is described. A password is normalized into a list of strings of a plurality of character types such as letters and numerals. For each string of letters, one or more corresponding letter permutations are determined which represent pronounceable combinations of that string. Then, for each letter permutation, a corresponding recognition grammar entry is created for a speech recognition grammar.

Abstract: For an audio coding, noise suppression is applied to an original audio signal to obtain an audio signal with reduced noise. A coding mode is selected based on the audio signal with reduced noise. The original audio signal is then encoded using this selected coding mode.

Abstract: The present invention automatically builds a contracted dictionary from a given list of multi-word proper names and performs fuzzy searches in the contracted dictionary. The contracted dictionary of proper names includes two linked trie-based dictionaries: a first dictionary is used to store single word names, each word name having an ID number; and a second dictionary is used to store multi-word names encoded with ID numbers. Information related to the multi-word names is also stored as a gloss to the terminal node of the multi-word entry of the trie-based dictionary. An approximate lookup for a multi-word name is conducted first for each word of the multi-word name using an approximate matching technique such as a phonetic proximity or a simple edit distance. Accordingly, N suggestions is determined for each word of the multi-word name under consideration. Then, multi-word candidates are assembled in ID notation.

Abstract: A system and method for phone detection. The system includes a microphone configured to receive a speech signal in an acoustic domain and convert the speech signal from the acoustic domain to an electrical domain, and a filter bank coupled to the microphone and configured to receive the converted speech signal and generate a plurality of channel speech signals corresponding to a plurality of channels respectively. Additionally, the system includes a plurality of onset enhancement devices configured to receive the plurality of channel speech signals and generate a plurality of onset enhanced signals. Each of the plurality of onset enhancement devices is configured to receive one of the plurality of channel speech signals, enhance one or more onsets of one or more signal pulses for the received one of the plurality of channel speech signals, and generate one of the plurality of onset enhanced signals.

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 dialog apparatus includes a dialog unit configured to perform a dialog with a user and to collect a plurality of items to be referred to in accordance with the dialog as the dialog continues, a dividing unit configured to divide the items into normal items related to any usable application and unusable items related to any usable application deleted, a managing unit configured to manage the items which the dialog unit refers to during the dialog with the user, in the normal items and the unusable items, an applying unit configured to apply a plurality of changes in the items to managing of the normal items and the unusable items, and a referring unit configured to refer to the managing unit in accordance with the items collected by the dialog unit and to determine to output a use-disapproval notice when the collected items include at least one unusable item.

Abstract: The invention provides a conversational speech analyzer which analyzes whether utterances in a meeting are of interest or concern. Frames are calculated using sound signals obtained from a microphone and a sensor, sensor signals are cut out for each frame, and by calculating the correlation between sensor signals for each frame, an interest level which represents the concern of an audience regarding utterances is calculated, and the meeting is analyzed.

Abstract: A sampling rate conversion apparatus and a method thereof are provided which increase the sampling rate of a discrete audio signal sampled at a predetermined sampling rate by using a fractal interpolation function (FIF). An audio signal portion formed by a predetermined number of sampling data items is divided into a plurality of interpolation intervals. On the audio signal portion, mapping points are determined. The number of the mapping points is in accordance with the degree of increase in the sampling rate. For the respective interpolation intervals, mapping parameters for performing mapping using the FIF on the mapping points are calculated. In all of the interpolation intervals, the mapping using the FIF is performed on the mapping points with the use of the mapping parameters according to the respective interpolation intervals. Thereby, new sampling data items are generated.

Abstract: A speech dialog control module enhances user operation of a speech dialog system by translating an input signal unrecognizable by a speech dialog system into a recognizable language. A speech dialog control module includes an input device that receives a speech signal in a first language. A controller receives the input signal and generates a control instruction that corresponds to the received input signal. The control instruction has a language that is different from the input signal. A speech-synthesis unit converts the control instruction into an output speech signal. An output device outputs the output speech signal.

Abstract: Systems and methods for voice activated commands in a digital home communication terminal are disclosed. One example method includes storing a program audio signal corresponding to a program tuned by the digital home communication terminal. The method also includes storing an incoming audio signal carrying speech and removing from the incoming audio signal a portion of the incoming audio signal that corresponds to the program audio signal, this producing an improved version of the incoming audio signal. The method also includes selecting one of a plurality of voice-activated commands that corresponds to the improved version of the incoming audio signal, and performing a function corresponding to the selected voice-activated command.

Abstract: A speech-to-text conversion system. The two-way speech recognition and dialect system comprises a computer system, an attached microphone assembly, and speech-to-text conversion software. The two-way speech recognition and dialect system includes a database of dialectal characteristics and queries a user to determine their likely dialect. The system uses this determination to reduce the time for the system to reliably transcribe a user's speech into text and to anticipate dialectal word usage. In another embodiment of the invention, the two-way speech recognition and dialect system is capable of transcribing the speech of multiple speakers while distinguishing between the different speakers and identifying the text belonging to each speaker.