Abstract: A method of testing the performance of real time DSP algorithms after customer code has been added and includes the steps of embedding a signature equation in the DSP code that calculates to a given signature value at the output of the DSP when running a test mode program through the DSP.

Abstract: A method and system of measuring layout efficiency is disclosed wherein after the initial layout (A), and the layout is drawn (B) a layout verification step C includes identifying seed devices or layers and the devices or layers are grown according to design rules and process rules to determine the minimum area required for the design. The layout verification is performed for both device packing density and interconnect packing density and the efficiency is calculated based on the total available area and reported.(D).

Abstract: A method for extracting wafer-to-wafer thickness variation from interferometry signals off patterned (product) wafer polish during non-endpointed CMP. The method includes sensing sample signals representing polishing trace from product wafers near the end of polishing period from at least two product wafers (101); estimating the value of the phase of the first and second wafers using polish data near the end of the polish period using nonlinear regression algorithm processing (103) and the GOF test (104); and calculating the difference in final thickness using the phase (105).

Abstract: A computer system includes a main processing unit (12) coupled to a DSP/memory module (40). The DSP/memory module (40) includes semiconductor memory (42) and digital signal processor circuitry (44) including one or more digital signal processors (56). The DSP/memory module (40) may be placed in standard main memory sockets, such as a SIMM or DIMM sockets, and used as conventional main memory. The memory module can also be used in a smart mode, wherein the digital signal processor (56) performs operations on data for retrieval by the main processing unit (12).

Abstract: A noise-dependent classifier for a speech recognition system includes a recognizer (15) that provides scores and score differences of two closest in-vocabulary words from a received utterance. A noise detector (17) detects the noise level of a pre-speech portion of the utterance. A classifier (19) is responsive to the detected noise level and scores and noise dependent model for making decisions for accepting or rejecting the utterance as a recognized word depending on the noise-dependent model and the scores.

Abstract: An improved method of implanting source and drain for CMOS devices is provided by a hard mask and dry etching to form polysilicon gates 20 percent longer than desired, implanting to form the source and drain of the PMOS transistor with dopant that moves faster during annealing such as Boron and then wet etching the polysilicon gates down to the shorter length such as the final length before implanting with the faster dopant such as arsenic.

Abstract: An improved method for mapping logical function test data of logical integrated circuits to physical representations uses a pruned diagnostic list. The steps include creating a final logical diagnostic list of potential bridging faults in response to testing the circuit for stuck-at faults at a plurality of nets of the circuit, receiving the physical data associated with nets of the circuit, applying adjacency criteria to the physical data, generating a pruned diagnostic list of potential bridging faults in response to applying the adjacency criteria, performing in-line inspection to obtain second localized probable defect data and correlating second localized portable defect data with the pruned diagnostic list.

Abstract: A collection of program instructions capable of executing on a host processor suitable for reading from a memory location of a target processor and suitable for creating a real-time data channel between said host and target processors.

Abstract: A Java television receiver (10) includes a Java television platform (12) that has a bidirectional Internet connection (14) capable of sending Internet data to the Java television platform (12) and transmitting data from the Java television platform (12) to an Internet provider. Additionally, a tuner/decoder (24) connected to the Java television platform (12) is provided which can receive, tune and decode television signals. An audio/video overlay (32) connected to the Java television platform is provided to combine displays from the Java television platform (12) and television signals from the tuner/decoder (24). Finally, a video output (36) and an audio output (38) sends the combined Internet data and television signals to a television (40).

Abstract: A method and system for generating and managing a knowledgebase for use in identifying anomalies on a manufactured object, such as a semiconductor wafer, includes measures for adding, deleting, and organizing data from the knowledgebase.

Abstract: A software predictive engine is described that provides a tool to analyze the import of defects in the software development process. By providing a prediction of escaping defects, in terms of the normalized sigma measure and the predicted lines of code, together with historical data the impact of defects at the stages of the design can be predicted. As actual defects are measured predicted sigma level can also be predicted. The engine includes generating a defect containment matrix for the historical data and for the current software program being developed.

Abstract: Speech recognition and the generation of speech recognition models is provided including the generation of unique phonotactic garbage models (15) to identify speech by, for example, English language constraints in addition to noise, silence and other non-speech models (11) and for speech recognition specific word models.

Abstract: A subframe-based correlation method for pitch and voicing is provided by finding the pitch track through a speech frame that minimizes pitch prediction residual energy over the frame. The method scans the range of possible time lags T and computes for each subframe within a given range of T the maximum correlation value and further finds the set of subframe lags to maximize the correlation over all of possible pitch lags.

Abstract: A method of and system for compressing and transmitting information which comprises recognizing a plurality of different types of information and providing a first representation of a recognized one of the plurality of different types of information having lower bandwidth transmission requirement than the recognized one of the plurality of different types of information. The first representation is transmitted to a remote receiver and a second representation of the recognized one of the recognized one of the plurality of different types of information at said remote receiver having greater information content than the first representation responsive to the first representation is provided. Responsive to the second representation, a recreation of the recognized one of the plurality of different types of information is provided. Recognizing includes providing a first data base and the second representation includes providing a second data base.

Abstract: A “multi-stage” method of estimating pitch in a speech encoder (FIG. 2). In a first stage of the method, a set of candidate pitch values is selected, such as by using a cost function that operates on said speech signal (steps 21-23). In a second stage of the method, a best candidate is selected. Specifically, in the second stage, pitch values calculated from previous speech segments are used to calculate an average pitch value (step 25). Then, depending on whether the average pitch value is short or long, one of two different analysis-by-synthesis (ABS) processes is then repeated for each candidate, such that for each iteration, a synthesized signal is derived from that pitch candidate and compared to a reference signal to provide an error value. A time domain ABS process is used if the average pitch is short (step 27), whereas a frequency domain ABS process is used if the average pitch is long (step 28).

Abstract: In a system (10) for adding speech recognition capabilities to Java, there is provided a speech recognition server (18) coupled to a Java application (24) through an application program interface (20). The Java application (24) dynamically specifies a grammar (56) to the application program interface (20), which communicates the grammar (56) to the speech recognition server (18). The speech recognition server (18) receives the grammar (56) and a speech input (66). Upon receiving the speech input (66), the speech recognition server (18) performs speech recognition and generates a result (58) based on the grammar (56). The application program interface (20) communicates the result (58) to the Java application (24), which performs an action based on the result (58) received.

Abstract: The search network in a speech recognition system is reduced by parsing the incoming speech expanding all active paths (101), comparing to speech models and scoring the paths and storing recognition level values at the slots (103) and accumulating the scores and discarding previous slots when a word end is detected creating a word end slot (109).

Abstract: A method and device for placement of sound sources in three-dimensional space via two loudspeakers. This technique uses an efficient implementation which consists of binaural signal processing and loudspeaker crosstalk cancellation, followed by panning into the left and right loudspeakers. For many applications, the binaural signal processing and crosstalk cancellation can be performed offline and stored in a file. Because, in this situation, panning is the only required operation, this technique results in a low-computation, real-time system for positional 3D audio over loudspeakers.

Abstract: A “multi-stage” method of estimating pitch in a speech encoder (FIG. 2). In a first stage of the method, a set of candidate pitch values is selected, such as by using a cost function that operates on said speech signal (steps 21-23). In a second stage of the method, a best candidate is selected. Specifically, in the second stage, pitch values calculated from previous speech segments are used to calculate an average pitch value (step 25). Then, depending on whether the average pitch value is short or long, one of two different analysis-by-synthesis (ABS) processes is then repeated for each candidate, such that for each iteration, a synthesized signal is derived from that pitch candidate and compared to a reference signal to provide an error value. A time domain ABS process is used if the average pitch is short (step 27), whereas a frequency domain ABS process is used if the average pitch is long (step 28).

Abstract: An apparatus and method for transmitting graphical representations is disclosed. The present invention has an image generator (112) that generates an image (114) of a head (106) with facial features, including an eye portion (202) that correlates to an eye (240) in the head (106). The apparatus also has a non-updating tracker (116) that searches for an eye target image of the eye (240) in the head (106). The eye portion (202) correlating to the eye (240) is positioned at the target image located by the non-updating tracker (116). The apparatus also has an updating tracker (118) that searches for an updated eye target image when the non-updating tracker (116) is unable to locate the eye target image. The updating tracker (118) also positions the eye portion (202) at the updated eye target image and updates the updated eye target image with the eye portion (202). The updating tracker further determines the location of the eye portion (202).