Abstract: A method for building a language model, a speech recognition method and an electronic apparatus are provided. The speech recognition method includes the following steps. Phonetic transcriptions of a speech signal are obtained from an acoustic model. Phonetic spellings matching the phonetic transcriptions are obtained according to the phonetic transcriptions and a syllable acoustic lexicon. According to the phonetic spellings, a plurality of text sequences and a plurality of text sequence probabilities are obtained from a language model. Each phonetic spelling is matched to a candidate sentence table; a word probability of each phonetic spelling matching a word in a sentence of the sentence table are obtained; and the word probabilities of the phonetic spellings are calculated so as to obtain the text sequence probabilities. The text sequence corresponding to a largest one of the sequence probabilities is selected as a recognition result of the speech signal.

Abstract: A method for building a language model, a speech recognition method and an electronic apparatus are provided. The speech recognition method includes the following steps. Phonetic transcriptions of a speech signal are obtained from an acoustic model. Phonetic spellings matching the phonetic transcriptions are obtained according to the phonetic transcriptions and a syllable acoustic lexicon. According to the phonetic spellings, a plurality of text sequences and a plurality of text sequence probabilities are obtained from a language model. Each phonetic spelling is matched to a candidate sentence table; a word probability of each phonetic spelling matching a word in a sentence of the sentence table are obtained; and the word probabilities of the phonetic spellings are calculated so as to obtain the text sequence probabilities. The text sequence corresponding to a largest one of the sequence probabilities is selected as a recognition result of the speech signal.

Abstract: A search method, a search system, and a natural language comprehension system are provided. The search system includes a structured database and a search engine. The structured database stores a plurality of records, each of which has a title field and a content field. The title field includes at least one sub-field, and each sub-field includes an indication field and a value field. The indication field stores indication data, the value field stores value data, and the content field stores detailed content data. The search engine conducts a full-text search to the records in the structured database according to a keyword derived from a user's request formation, and a search result is transmitted to a knowledge comprehension assistance module, so as to recognize the user's intention. After the user's intention is recognized, information associated with the recognized user's intention is transmitted back to the user.

Abstract: A search method, a search system, and a natural language comprehension system are provided. The search system includes a structured database and a search engine. The structured database stores a plurality of records, each of which has a title field and a content field. The title field includes at least one sub-field, and each sub-field includes an indication field and a value field. The indication field stores indication data, the value field stores value data, and the content field stores detailed content data. The search engine conducts a full-text search to the records in the structured database according to a keyword derived from a user's request formation, and a search result is transmitted to a knowledge comprehension assistance module, so as to recognize the user's intention. After the user's intention is recognized, information associated with the recognized user's intention is transmitted back to the user.

Abstract: One or more processors migrate an amount of cloud data from a non-volatile memory to a volatile cache memory. One or more processors partition the amount of cloud data into a plurality of objects of a size that allow for a satisfactory read/write throughput performance. One or more processors analyze the plurality of objects for input/output performance over a time period. One or more processors migrate a first portion of the plurality of objects back to the non-volatile memory, wherein the first portion of the plurality of objects exhibit a first input/output performance lower than a first threshold value.

Abstract: A speech recognition method and an electronic apparatus are provided. The speech recognition method includes the following steps. A plurality of phonetic transcriptions of a speech signal is obtained according to an acoustic model. A phonetic spelling and intonation information matched to the phonetic transcriptions are obtained according to a phonetic transcription sequence and a syllable acoustic lexicon of the invention. According to the phonetic spellings and the intonation information, a plurality of phonetic spelling sequences and a plurality of phonetic spelling sequence probabilities are obtained from a language model. The phonetic spelling sequence corresponding to a largest one among the phonetic spelling sequence probabilities is selected as a recognition result of the speech signal.

Abstract: A sorting method of data documents is provided, adapted to an electronic device. The sort method includes the following steps: retrieving a plurality of keywords from contents of a plurality of data documents; retrieving a keyword ranking corresponding to the at least one first keyword by a search engine; searching a keyword category corresponding to the at least one first keyword; and inputting the at least one first keyword, the keyword ranking and the keyword category of each of the at least one first keyword into a sort algorithm thereby outputting a predicting ranking of the first data document to sort the first data document, wherein the sort algorithm is generated based on contents of a plurality of second data documents and a current ranking of each of the plurality of second data documents.

Abstract: A sorting method of data documents is provided, adapted to an electronic device. The sort method includes the following steps: retrieving a plurality of keywords from contents of a plurality of data documents; retrieving corresponding keyword rankings of the plurality of keywords by a search engine; searching corresponding keyword categories of the plurality of keywords; and generating a sort algorithm based on the plurality of keywords, the keyword ranking and the keyword category of each of the plurality of keywords, and a current ranking of each of the plurality of data documents, wherein the sort algorithm is used to calculate a predicting ranking of another data document and to sort the another data document.

Abstract: A wafer process for molded chip scale package (MCSP) comprises: depositing metal bumps on bonding pads of chips on a wafer; forming a first packaging layer at a front surface of the wafer to cover the metal bumps; forming an un-covered ring at an edge of the wafer to expose two ends of each scribe line of a plurality of scribe lines; thinning the first packaging layer to expose metal bumps; forming cutting grooves; grinding a back surface of the wafer to form a recessed space and a support ring at the edge of the wafer; depositing a metal seed layer at a bottom surface of the wafer in the recessed space; cutting off an edge portion of the wafer; flipping and mounting the wafer on a substrate; depositing a metal layer covering the metal seed layer; removing the substrate from the wafer; and separating individual chips from the wafer by cutting through the first packaging layer, the wafer, the metal seed layers and the metal layers along the scribe lines.

Abstract: A method for building a language model, a speech recognition method and an electronic apparatus are provided. The speech recognition method includes the following steps. Phonetic transcriptions of a speech signal are obtained from an acoustic model. Phonetic spellings matching the phonetic transcriptions are obtained according to the phonetic transcriptions and a syllable acoustic lexicon. According to the phonetic spellings, a plurality of text sequences and a plurality of text sequence probabilities are obtained from a language model. Each phonetic spelling is matched to a candidate sentence table; a word probability of each phonetic spelling matching a word in a sentence of the sentence table are obtained; and the word probabilities of the phonetic spellings are calculated so as to obtain the text sequence probabilities. The text sequence corresponding to a largest one of the sequence probabilities is selected as a recognition result of the speech signal.

Abstract: A natural language dialog system and a method capable of correcting a speech response are provided. The method includes following steps. A first speech input is received. At least one keyword included in the first speech input is parsed to obtain a candidate list having at least one report answers. One of the report answers is selected from the candidate list as a first report answer, and a first speech response is output according to the first report answer. A second speech input is received and parsed to determine whether the first report answer is correct. If the first report answer is incorrect, another report answer other than the first report answer is selected from the candidate list as a second report answer. According to the second report answer, a second speech response is output.

Abstract: A control method of an electronic device with an image capture function is provided. The control method includes the following steps: detecting a distance between the electronic device and at least one subject in a scene to be captured to generate a distance detection result; and utilizing a controller for referring to the distance detection result to determine whether to turn on or turn off a flashlight of the electronic device to capture the scene.

Abstract: Roughly described, a system and method for delivering video content to a user's client device in a video-on-demand (VOD) system, which includes providing a collection of video segments, the segments having a predefined default sequence; establishing a streaming video session according to a session-oriented protocol; transmitting toward the client device a script executable by the client device, the script operable to transmit navigational codes toward the head-end equipment in response to and indicating user selection among navigational choices; beginning transmission of the video segments in the collection toward the client device in accordance with the default sequence of segments; and in response to receipt of one of the navigational codes, and without tearing down the streaming video session, altering the transmission sequence to jump to the segment that the user selected.

Abstract: A wafer process for molded chip scale package (MCSP) comprises: depositing metal bumps on bonding pads of chips on a wafer; forming a first packaging layer at a front surface of the wafer to cover the metal bumps; forming an un-covered ring at an edge of the wafer to expose two ends of each scribe line of a plurality of scribe lines; thinning the first packaging layer to expose metal bumps; forming cutting grooves; grinding a back surface of the wafer to form a recessed space and a support ring at the edge of the wafer; depositing a metal seed layer at a bottom surface of the wafer in the recessed space; cutting off an edge portion of the wafer; flipping and mounting the wafer on a substrate; depositing a metal layer covering the metal seed layer; removing the substrate from the wafer; and separating individual chips from the wafer by cutting through the first packaging layer, the wafer, the metal seed layers and the metal layers along the scribe lines.

Abstract: A wafer process for MCSP comprises: depositing a metal bump on bonding pads of chips; forming a first packaging layer at front surface of wafer covering metal bumps while forming an un-covered ring at the edge of wafer to expose the ends of each scribe line located between two adjacent chips; thinning first packaging layer to expose metal bumps; grinding back surface of wafer to form a recessed space and a support ring at the edge of the wafer; depositing a metal seed layer and a thick metal layer at bottom surface of wafer in recessed space in a sequence; cutting off the edge portion of wafer; and separating individual chips from wafer by cutting through first packaging layer, the wafer and the metal seed and metal layers along the scribe line.

Abstract: The present invention discloses a method for deduplication of a file, a computer program product, and an apparatus thereof. In the method, the file is partitioned into at least one composite block, wherein the composite block includes a fixed-size block and a variable-size block, the variable-size block being determined based on content of the file. Then a deduplication operation is performed on the at least one composite block.

Abstract: Roughly described, a system and method for delivering video content to a user's client device in a video-on-demand (VOD) system, which includes providing a collection of video segments, the segments having a predefined default sequence; establishing a streaming video session according to a session-oriented protocol; transmitting toward the client device a script executable by the client device, the script operable to transmit navigational codes toward the head-end equipment in response to and indicating user selection among navigational choices; beginning transmission of the video segments in the collection toward the client device in accordance with the default sequence of segments; and in response to receipt of one of the navigational codes, and without tearing down the streaming video session, altering the transmission sequence to jump to the segment that the user selected.

Abstract: A method and system for controlling multiple displays is provided. The disclosed method is used to control a plurality of graphics processing units (GPUs), wherein every GPU controls one or more displays. The method includes the following steps: providing a graphical interface the same to a graphical program library of an operating system to replace the graphical program library to receive a drawing command from an application program; determining a display set of the GPUs according to a display region of the application program, wherein a frame displayed by the display controlled by each GPU is intersected to the display region; and delivering coordinate-transformed drawing commands to the GPUs in the display set according to the display intersection region, wherein each GPU in the display set only draws the content of the corresponding display intersection region.

Abstract: A voice control device and a corresponding voice control method are provided. The voice control device includes a sound receiver, a sound converter, a voice identifier, and a central processing unit (CPU). The sound receiver receives a first sound signal. The sound converter converts the first sound signal from analog signal to digital signal. The voice identifier identifies a first voice signal from the first sound signal, performs a first comparison on the first voice signal and a second voice signal, and generates a wake-up signal according to the first comparison. When receiving the wake-up signal, the CPU enters a working state from a sleeping state, performs a second comparison on the first voice signal and the second voice signal, and takes over the voice input from the sound receiver and the sound converter according to the second comparison.