Abstract: A Viterbi trellis processing technique in which soft decisions and hard decisions are derived from a received signal and the soft decisions are enhanced by being modified using the hard decisions. A log likelihood ratio for a bit of the received signal can be derived by grouping candidate metrics associated with the decision that the bit has a first state, grouping candidate metrics associated with the decision that the bit has a second state, applying respective functions to the groups and calculating the difference of the function values.

Abstract: A Viterbi decoder includes: an observation vector sequence generator for generating an observation vector sequence by converting an input speech to a sequence of observation vectors; a local optimal state calculator for obtaining a partial state sequence having a maximum similarity up to a current observation vector as an optimal state; an observation probability calculator for obtaining, as a current observation probability, a probability for observing the current observation vector in the optimal state; a buffer for storing therein a specific number of previous observation probabilities; a non-linear filter for calculating a filtered probability by using the previous observation probabilities stored in the buffer and the current observation probability; and a maximum likelihood calculator for calculating a partial maximum likelihood by using the filtered probability.

Abstract: Systems and methods are provided for generating error events for decoded bits using a Soft output Viterbi algorithm (SOYA). A winning path through a trellis can be determined and decoded information can be generated. Path metric differences can be computed within the trellis based on the winning path. A plurality of error event masks and error event metrics can be generated based on the decoded information and the path metric differences.

Abstract: A multi-channel sequential Viterbi decoder includes: an input data buffer, a “Read Single Data Word from Input Data Buffer” signal driver, a processing unit selector, a decoder channel parameters registers unit, a processing unit for the “Reset Path Metrics” command, a processing unit for the “Set Path Metric Value for the Given Path Number” command, a processing unit for the “Get Single Bit from the Path with Given Number” command, a processing unit for the “Process Input Samples” command, a decoding paths and path metrics RAM, a unit for generating current decoder channel base address for the decoding paths and path metrics RAM, a unit for generating cell address for the decoding path and path metric RAM, and a data buffers unit for decoder channels output.

Abstract: A Viterbi pack instruction is disclosed that masks the contents of a first predicate register with a first masking value and masks the contents of a second predicate register with a second masking value. The resulting masked data is written to a destination register. The Viterbi pack instruction may be implemented in hardware, firmware, software, or any combination thereof.

Abstract: Repetition of content words in a communication is used to increase the certainty, or, alternatively, reduce the uncertainty, that the content words were actual words from the communication. Reducing the uncertainty of a particular content word of a communication in turn increases the likelihood that the content word is relevant to the communication. Reliable, relevant content words mined from a communication can be used for, e.g., automatic internet searches for documents and/or web sites pertinent to the communication. Reliable, relevant content words mined from a communication can also, or alternatively, be used to automatically generate one or more documents from the communication, e.g., communication summaries, communication outlines, etc.

Abstract: A method for classifying data includes selecting an elemental size and features for the data that are representative of possible subclasses. Resolution widths are selected in conjunction with these features. Models associated with symbols are developed from these resolution widths and features. Data is compared with these models to give a likelihood that the model applies. The best model is determined and a signal is provided related to the symbol associated with the best model.

Abstract: A method for communication includes receiving a spatially-multiplexed signal using multiple receivers to produce multiple respective received signals. The spatially-multiplexed signal includes multiple simultaneously-transmitted symbols, which are selected from respective sets of constellation symbols, each constellation symbol representing a respective set of values of a group of data bits. Combinations of the constellation symbols are traversed iteratively. Each combination includes one constellation symbol from each of the sets of the constellation symbols and represents N data bits. The traversed combinations are searched for a combination that matches the received signals. During traversal of the combinations, at least 2N measures of likelihood regarding the values of the data bits represented by each traversed combination are accumulated. The accumulated measures of likelihood are processed to produce soft bit metrics.

Abstract: Disclosed herein is a system, method and computer-readable medium storing instructions for controlling a computing device according to the method. The invention relates to a system, method and computer-readable medium storing instructions for controlling a computing device according to the method. As an example embodiment, the method uses a speech recognition decoder that operates or uses fixed point arithmetic. The exemplary method comprises representing arc costs associated with at least one finite state transducer (FST) in fixed point, representing parameters associated with a hidden Markov model (HMM) in fixed point and processing speech data in the speech recognition decoder using fixed point arithmetic for the fixed point FST arc costs and the fixed point HMM parameters. The method may also include computing at the decoder sentence hypothesis probabilities with fixed point arithmetic as type Q-2e numbers.

Abstract: An information communication terminal (100) that includes: a speech recognition module (6) for recognizing speech information to identify a plurality of words in the recognized speech information; a storage medium (20) for storing keyword extraction condition setting data (24) in which a condition for extracting a keyword is set; a keyword extraction module (8) for reading the keyword extraction condition setting data (24) to extract a plurality of keywords from the plurality of words; a related information acquisition module (11) for acquiring related information related to a plurality of keywords; and a related information output module (14) for providing related information to a monitor (2).

Abstract: Methods, apparatus, and products are disclosed for adjusting a speech engine for a mobile computing device based on background noise, the mobile computing device operatively coupled to a microphone, that include: sampling, through the microphone, background noise for a plurality of operating environments in which the mobile computing device operates; generating, for each operating environment, a noise model in dependence upon the sampled background noise for that operating environment; and configuring the speech engine for the mobile computing device with the noise model for the operating environment in which the mobile computing device currently operates.

Abstract: An information processing apparatus includes a plurality of information input units that inputs observation information of a real space, an event detection unit that generates event information including estimated position information and estimated identification (ID) information of a user present in the real space based on analysis of the information input from the information input unit, and an information integration processing unit that inputs the event information, and generates target information including a position and user ID information of each user based on the input event information and signal information representing a probability value for an event generating source. Here, the information integration processing unit includes an utterance source probability calculation unit having an identifier, and calculates an utterance source probability based on input information using the identifier in the utterance source probability calculation unit.

Abstract: A recognition (e.g., speech, handwriting, etc.) model build process that is declarative and data-dependence-based. Process steps are defined in a declarative language as individual processors having input/output data relationships and data dependencies of predecessors and subsequent process steps. A compiler is utilized to generate the model building sequence. The compiler uses the input data and output data files of each model build processor to determine the sequence of model building and automatically orders the processing steps based on the declared input/output relationship (the user does not need to determine the order of execution). The compiler also automatically detects ill-defined processes, including cyclic definition and data being produced by more than one action. The user can add, change and/or modify a process by editing a declaration file, and rerunning the compiler, thereby a new process is automatically generated.

Abstract: A speech recognition circuit comprises an input buffer for receiving processed speech parameters. A lexical memory contains lexical data for word recognition. The lexical data comprises a plurality of lexical tree data structures. Each lexical tree data structure comprises a model of words having common prefix components. An initial component of each lexical tree structure is unique. A plurality of lexical tree processors are connected in parallel to the input buffer for processing the speech parameters in parallel to perform parallel lexical tree processing for word recognition by accessing the lexical data in the lexical memory. A results memory is connected to the lexical tree processors for storing processing results from the lexical tree processors and lexical tree identifiers to identify lexical trees to be processed by the lexical tree processors.

Abstract: The present invention discloses a method for training an exception-limited phonetic decision tree. An initial subset of data can be selected and used for creating an initial phonetic decision tree. Additional terms can then be incorporated into the subset. The enlarged subset can be used to evaluate the phonetic decision tree with the results being categorized as either correctly or incorrectly phonetized. An exception-limited phonetic tree can be generated from the set of correctly phonetized terms. If the termination conditions for the method have been determined to be unsatisfactorily met, then steps of the method can be repeated.

Abstract: An optimal selection or decision strategy is described through an example that includes use in dialog systems. The selection strategy or method includes receiving multiple predictions and multiple probabilities. The received predictions predict the content of a received input and each of the probabilities corresponds to one of the predictions. In an example dialog system, the received input includes an utterance. The selection method includes dynamically selecting a set of predictions from the received predictions by generating ranked predictions. The ranked predictions are generated by ordering the plurality of predictions according to descending probability.

Abstract: A method, system and communication device for enabling uniterm discovery from audio content and voice-to-voice searching of audio content stored on a device using discovered uniterms. Received audio/voice input signal is sent to a uniterm discovery and search (UDS) engine within the device. The audio data may be associated with other content that is also stored within the device. The UDS engine retrieves a number of uniterms from the audio data and associates the uniterms with the stored content. When a voice search is initiated at the device, the UDS engine generates a statistical latent lattice model from the voice query and scores the uniterms from the audio database against the latent lattice model. Following a further refinement, the best group of uniterms is then determined and segments of the stored audio data and/or other content corresponding to the best group of uniterms are outputted.

Abstract: Improved audio source separation is provided by providing an audio dictionary for each source to be separated. Thus the invention can be regarded as providing “partially blind” source separation as opposed to the more commonly considered “blind” source separation problem, where no prior information about the sources is given. The audio dictionaries are probabilistic source models, and can be derived from training data from the sources to be separated, or from similar sources. Thus a library of audio dictionaries can be developed to aid in source separation. An unmixing and deconvolutive transformation can be inferred by maximum likelihood (ML) given the received signals and the selected audio dictionaries as input to the ML calculation. Optionally, frequency-domain filtering of the separated signal estimates can be performed prior to reconstructing the time-domain separated signal estimates. Such filtering can be regarded as providing an “audio skin” for a recovered signal.

Abstract: Included are embodiments for providing speech analysis. At least one embodiment of a method includes receiving audio data associated with a communication and providing the at least one phoneme in a phonetic transcript, the phonetic transcript including at least one character from a phonetic alphabet.

Abstract: According to an embodiment, voice recognition apparatus includes units of: acoustic processing, voice interval detecting, dictionary, collating, search target selecting, storing and determining, and voice recognition method includes processes of: selecting a search range on basis of a beam search, setting and storing a standard frame, storing an output probability of a certain transition path, determining whether or not the output probability of a certain path is stored. Number of times of calculation of the output probability is reduced by selecting the search range on basis of the beam search, calculating the output probability of the certain transition path only once in an interval from when the standard frame is set to when the standard frame is renewed, and storing and using thus calculated value as an approximate value of the output probability in subsequent frames.

Abstract: A time-synchronous lattice-constrained search algorithm is developed and used to process a linguistic model of speech that has a long-contextual-span capability. In the algorithm, hypotheses are represented as traces that include an indication of a current frame, previous frames and future frames. Each frame can include an associated linguistic unit such as a phone or units that are derived from a phone. Additionally, pruning strategies can be applied to speed up the search. Further, word-ending recombination methods are developed to speed up the computation. These methods can effectively deal with an exponentially increased search space.

Abstract: According to some embodiments, a method and apparatus are provided to buffer N audio frames of a plurality of audio frames associated with an audio signal, pre-compute scores for a subset of context dependent models (CDMs), and perform a graphical model search associated with the N audio frames where a score of a context independent model (CIM) associated with a CDM is used in lieu of a score for the CDM when a score for the CDM is needed and has not been pre-computed.

Abstract: The present invention discloses a method for semantically processing speech for speech recognition purposes. The method can reduce an amount of memory required for a Viterbi search of an N-gram language model having a value of N greater than two and also having at least one embedded grammar that appears in a multiple contexts to a memory size of approximately a bigram model search space with respect to the embedded grammar. The method also reduces needed CPU requirements. Achieved reductions can be accomplished by representing the embedded grammar as a recursive transition network (RTN), where only one instance of the recursive transition network is used for the contexts. Other than the embedded grammars, a Hidden Markov Model (HMM) strategy can be used for the search space.

Abstract: A phrase-based translation system and method includes a statistically integrated phrase lattice (SIPL) (H) which represents an entire translational model. An input (I) is translated by determining a best path through an entire lattice (S) by performing an efficient composition operation between the input and the SIPL. The efficient composition operation is performed by a multiple level search where each operand in the efficient composition operation represents a different search level.

Abstract: A receiver for use in a wireless network comprising a communications channel and a method of allocating deinterleaver memory usage in the receiver, wherein the receiver comprises a processor adapted to organize subchannels of the communications channel and set a number (N) of data bits per soft decision, wherein the soft decision is represented by N data bits; an address decoder adapted to decode the subchannels; a demapper adapted to receive QAM symbols and demap the QAM symbols to soft decisions; a deinterleaver adapted to perform deinterleaving on the soft decisions, wherein the deinterleaver comprises a memory component having a storage size that is a function of the number (N) of bits per soft decision; and a Viterbi decoder adapted to decode the deinterleaved soft decisions.

Abstract: A method for recognizing a speaker of an utterance in a speech recognition system is disclosed. A likelihood score for each of a plurality of speaker models for different speakers is determined. The likelihood score indicating how well the speaker model corresponds to the utterance. For each of the plurality of speaker models, a probability that the utterance originates from that speaker is determined. The probability is determined based on the likelihood score for the speaker model and requires the estimation of a distribution of likelihood scores expected based at least in part on the training state of the speaker.

Abstract: A method of encoding data for transmission to one or more users selects a given number of bits of data from a transport block to be subject to hybrid ARQ functionality for channel coding. Only the selected bits are channel coded in a HARQ block for subsequent transmission using a given set of channelization codes to one or more users.

Abstract: A MIMO receiver implements a method for the soft bit metric calculation with linear MIMO detection for LDPC codes, after linear matrix inversion MIMO detection. In the receiver, a detector detects the estimated symbol and the noise variance. Further, a soft metric calculation unit computes the distance between the estimated symbol and the constellation point, and then divides the distance by the noise variance to determine the soft bit metrics.

Abstract: A Viterbi decoder includes: an observation vector sequence generator for generating an observation vector sequence by converting an input speech to a sequence of observation vectors; a local optimal state calculator for obtaining a partial state sequence having a maximum similarity up to a current observation vector as an optimal state; an observation probability calculator for obtaining, as a current observation probability, a probability for observing the current observation vector in the optimal state; a buffer for storing therein a specific number of previous observation probabilities; a non-linear filter for calculating a filtered probability by using the previous observation probabilities stored in the buffer and the current observation probability; and a maximum likelihood calculator for calculating a partial maximum likelihood by using the filtered probability.

Abstract: A quadrature amplitude modulation trellis coded modulation (QAM-TCM) decoding apparatus and the related method that receives and decodes a QAM signal. The QAM-TCM decoding apparatus includes an in-phase least significant bit (LSB) decoding path, which includes a in-phase Viterbi decoder for executing a decoding procedure on at least one LSB corresponding to an in-phase component of the QAM signal, a quadrature-phase LSB decoding path, which includes a quadrature-phase Viterbi decoder for executing a decoding procedure on at least one LSB corresponding to a quadrature-phase component of the QAM signal, and a most significant bit (MSB) decoding path for executing a decoding procedure on MSB portions corresponding to the in-phase or the quadrature-phase of the QAM signal.

Abstract: A pattern matching method for matching between a first symbol sequence and a second symbol sequence which is shorter than the first symbol sequence is provided. The method includes the steps of performing DP matching between the first and second symbol sequences to create a matrix of the DP matching transition, detecting the maximum length of lengths of consecutive correct answers based on the matrix of the DP matching transition, and calculating similarity based on the maximum length.

Abstract: A path memory circuit for use in a Viterbi decoding process performed based on state transitions through a number n (n is a positive integer) of states. The path memory circuit includes a memory area A formed by the storage circuits of the first to ith (i is an integer from 0 to M) stages; a memory area B formed by the selective storage circuits that select and hold a decoding result for any state k (k is integer from 1 to n) of the storage circuits from the i+1th stage to the Mth stage; and a memory area C formed by the selective storage circuits other than the memory area A and the memory area B.

Abstract: Method for online character recognition of Arabic text, the method including receiving handwritten Arabic text from a user in the form of handwriting strokes, sampling the handwriting strokes to acquire a sequence of two dimensional point representations thereof, with associated temporal data, geometrically pre processing and extracting features on the point representations, detecting delayed strokes and word parts in the pre processed point representations, projecting the delayed strokes onto the body of the word parts, constructing feature vector representations for each word part, thereby generating an observation sequence, and determining the word with maximum probability given the observation sequence, resulting in a list of word probabilities.

Abstract: The present invention relates to speech recognition systems, particularly speech-to-text systems and software and decoders for the same.

Abstract: In some speech recognition applications, not only the language of the utterance is not known in advance, but also a single utterance may contain words in more than one language. At the same time, it is impractical to build speech recognizers for all expected combinations of languages. Moreover, business needs may require a new combination of languages to be supported in short period of time. The invention addresses this issue by a novel way of combining and controlling the components of the single-language speech recognizers to produce multilingual speech recognition functionality capable of recognizing multilingual utterances at a modest increase of computational complexity.

Abstract: In a radio communication system, transmitter and receiver stations share information on a maximum number of bits communicated per symbol. The transmitter station encodes a signal with sufficient error correcting capabilities to create a codeword. The transmitter station allocates the bits from the codeword to each symbol, modulates the symbols using a modulation type which processes symbols each having a number of bits equal to or smaller than the maximum number of bits per symbol, and transmits the modulated symbols. The receiver station demodulates the symbols using a modulation type which processes a larger number of bits per symbol as the transmission path quality is higher from among modulation types which process symbols having a number of bits equal to or smaller than the maximum number of bits per symbol.

Abstract: A method and system for decoding WCDMA AMR speech data using redundancy may include generating at least one bit-sequence for at least one of a plurality of channels that comprises received WCDMA speech data. The bit-sequence may be generated by using a decoding algorithm and may be decrypted to recover the data that may have been encrypted before being transmitted. At least one bit-sequence may be selected for each of the channels by using redundancy, such as, for example, CRC, in the received WCDMA speech data. The redundancy in the received WCDMA speech data may be, for example, CRC. The bit-sequence for each of the channels may be combined to form at least one speech stream. A speech stream may be selected based on speech constraints, which may comprise gain continuity and/or pitch continuity. The selected speech stream may be communicated to a voice decoder.

Abstract: A speech recognition circuit comprises an input buffer for receiving processed speech parameters. A lexical memory contains lexical data for word recognition. The lexical data comprises a plurality of lexical tree data structures. Each lexical tree data structure comprises a model of words having common prefix components. An initial component of each lexical tree structure is unique. A plurality of lexical tree processors are connected in parallel to the input buffer for processing the speech parameters in parallel to perform parallel lexical tree processing for word recognition by accessing the lexical data in the lexical memory. A results memory is connected to the lexical tree processors for storing processing results from the lexical tree processors and lexical tree identifiers to identify lexical trees to be processed by the lexical tree processors.

Abstract: A speech recognition method including: layering a central lexicon in a tree structure with respect to recognition-subject vocabularies; performing multi-pass symbol matching between a recognized phoneme sequence and a phonetic sequence of the central lexicon layered in the tree structure; and selecting a final speech recognition result via a Viterbi search process using a detailed acoustic model with respect to candidate vocabularies selected by the multi-pass symbol matching.

Abstract: Systems and modules for use in trellis-based decoding of convolutionally encoded sets of data bits. A first calculation module receives an encoded set of data bits and calculates a signal distance or a measure of the differences between the encoded set and each one of a group of predetermined states. The first calculation module consists of multiple parallel calculation submodules with each submodule being tasked to perform an XOR operation between the encoded set and one of the predetermined states. Multiple parallel second calculation modules each receiving the output of the first calculation module, calculate cumulative signal distances using the output of the first calculation module. Each second calculation module outputs its lowest valued cumulative signal distance and this may be used as input to a memory system for storing a database used in further decoding of the encoded data.

Abstract: A method and apparatus for performing channel compensation and symbol demodulation using an estimated channel impulse response during coherent demodulation of a received Orthogonal Frequency Division Multiplexing (OFDM) signal are provided. In the apparatus, an FFT processor IFFT-processes a received signal. A channel compensator generates a channel-compensated signal by multiplying the FFT received signal by an estimated channel impulse response and calculates the power of the estimated channel impulse response. A symbol demodulator sets the power of the estimated channel impulse response as a reference point defining a minimum distance between signal points in a signal constellation, and decides soft metric values for channel decoding using the reference point and I-channel and Q-channel signal components of the channel-compensated signal. A decoder recovers information bits by decoding the soft metric values.

Abstract: VoIP phones according to the present invention include a microphone, which may be internal or external, and allow the user to communicate unobtrusively, check voice mail and conduct other activities in an environment which can be noisy in general and extremely noisy sometimes. Speech recognition functionally may also be used to generate and send touch tone or DTMF tones such as in response to call trees or voice recognition functionality used by airlines, credit card companies, voice mail systems, and other applications. A system and method of audio processing which provides enhanced speech recognition is provided. Audio input is received at the microphone which is processed by adaptive noise cancellation to generate an enhanced audio signal. The operation of the speech recognition engine and the adaptive noise canceller may be advantageously controlled based on Voice Activity Detection (VAD).

Abstract: This invention modifies Viterbi decoding to improve BER. Within the state metric unit cascade block, this invention forces the unused ACS units decision bits to a 0 for the top rail and a 1 for the bottom rail. This invention modifies the final maximum state index with the selected decision bits from the unused ACS units. This invention uses the modified final maximum state index as the initial conditions for the k?1 traceback shift register. This invention also uses the final maximum state index to mask the generated pretraceback decision bits generated from the last block of ACS units.

Abstract: A speech recognition method and apparatus perform speaker clustering and speaker adaptation using average model variation information over speakers while analyzing the quantity variation amount and the directional variation amount. In the speaker clustering method, a speaker group model variation is generated based on the model variation between a speaker-independent model and a training speaker ML model. In the speaker adaptation method, the model in which the model variation between a test speaker ML model and a speaker group ML model to which the test speaker belongs which is most similar to a training speaker group model variation is found, and speaker adaptation is performed on the found model. Herein, the model variation in the speaker clustering and the speaker adaptation are calculated while analyzing both the quantity variation amount and the directional variation amount. The present invention may be applied to any speaker adaptation algorithm of MLLR and MAP.

Abstract: The present invention relates to a method for detecting a signal in an MIMO system. In the method, a received signal is detected in a zero forcing (ZF) method, and a first detection interval is established from the signal detected in the ZF method. The received signal is detected within the first detection interval in a maximum likelihood (ML) method, a second detection interval is established from the signals respectively detected in the ZF method and the ML method. A final solution is determined by detecting the received signal within the second detection interval in the ML method.

Abstract: A speech recognition circuit comprises an input buffer for receiving processed speech parameters. A lexical memory contains lexical data for word recognition. The lexical data comprises a plurality of lexical tree data structures. Each lexical tree data structure comprises a model of words having common prefix components. An initial component of each lexical tree structure is unique. A plurality of lexical tree processors are connected in parallel to the input buffer for processing the speech parameters in parallel to perform parallel lexical tree processing for word recognition by accessing the lexical data in the lexical memory. A results memory is connected to the lexical tree processors for storing processing results from the lexical tree processors and lexical tree identifiers to identify lexical trees to be processed by the lexical tree processors.

Abstract: Building a language model for use in speech recognition includes identifying without user interaction a source of text related to a user. Text is retrieved from the identified source of text and a language model related to the user is built from the retrieved text.

Abstract: Methods, systems, and apparatus are provided to generate a Viterbi path for a DBN. The DBN is converted to a chain of junction trees, where each tree represents a decision-making process. The trees are forwardly iterated and the Viterbi path is generated during the forward iteration (forward pass). This is achieved by maintaining backpointers to previously processed junction trees during the forward pass and dynamically assembling the Viterbi with each pair of junction trees during the forward pass.

Abstract: A method of identifying a location of a query string in an audio signal is provided. Under the method, a segment of the audio signal is selected. A score for a query string in the segment of the audio signal is determined by determining the product of probabilities of overlapping sequences of tokens. The score is then used to decide if the segment of the audio signal is likely to contain the query string.

Abstract: The base station transceiver in a CDMA mobile communication system is configured to have a separate hardware component for performing only a Viterbi decoding apart from a single hardware H/W that performs a composite function of modulation/demodulation and Viterbi encoding/decoding. The modulator and the Viterbi encoder are provided in one hardware by sectors; more than one demodulator being provided by sectors for demodulating signals from multiple users. And, more than one Viterbi decoder is separately provided for performing a Viterbi decoding of the signals demodulated at the plural demodulator constituted in each sector, thereby facilitating a decoding of demodulated signals received from multiple users and enhancing efficiency of the hardware.