A method includes receiving audio data encoding an utterance, processing, using a speech recognition model, the audio data to generate speech recognition scores for speech elements, and determining context scores for the speech elements based on context data indicating a context for the utterance. The method also includes executing, using the speech recognition scores and the context scores, a beam search decoding process to determine one or more candidate transcriptions for the utterance. The method also includes selecting a transcription for the utterance from the one or more candidate transcriptions.

In some embodiments, recognition results produced by a speech processing system (which may include two or more recognition results, including a top recognition result and one or more alternative recognition results) based on an analysis of a speech input, are evaluated for indications of potential errors. In some embodiments, the indications of potential errors may include discrepancies between recognition results that are meaningful for a domain, such as medically-meaningful discrepancies. The evaluation of the recognition results may be carried out using any suitable criteria, including one or more criteria that differ from criteria used by an ASR system in determining the top recognition result and the alternative recognition results from the speech input. In some embodiments, a recognition result may additionally or alternatively be processed to determine whether the recognition result includes a word or phrase that is unlikely to appear in a domain to which speech input relates.

A system includes acquisition of meeting data associated with a meeting, determination of a plurality of meeting participants based on the acquired meeting data, acquisition of e-mail data associated with each of the plurality of meeting participants, generation of a meeting language model based on the acquired e-mail data and the meeting data, and transcription of audio associated with the meeting based on the meeting language model.

A method of performing speech recognition using a two-pass deliberation architecture includes receiving a first-pass hypothesis and an encoded acoustic frame and encoding the first-pass hypothesis at a hypothesis encoder. The first-pass hypothesis is generated by a recurrent neural network (RNN) decoder model for the encoded acoustic frame. The method also includes generating, using a first attention mechanism attending to the encoded acoustic frame, a first context vector, and generating, using a second attention mechanism attending to the encoded first-pass hypothesis, a second context vector. The method also includes decoding the first context vector and the second context vector at a context vector decoder to form a second-pass hypothesis

A system includes acquisition of meeting data associated with a meeting, determination of a plurality of meeting participants based on the acquired meeting data, acquisition of e-mail data associated with each of the plurality of meeting participants, generation of a meeting language model based on the acquired e-mail data and the meeting data, and transcription of audio associated with the meeting based on the meeting language model.

An interactive speech recognition system includes a database containing a plurality of reference terms, a list memory that receives the reference terms of category “n,” a processing circuit that populates the list memory with the reference terms corresponding to the category “n,” and a recognition circuit that processes the reference terms and terms of a spoken phrase. The recognition circuit determines if a reference term of category “n” matches a term of the spoken phrase.

A system and method are provided for performing speech processing. A system includes an audio detection system configured to receive a signal including speech and a memory having stored therein a database of keyword models forming an ensemble of filters associated with each keyword in the database. A processor is configured to receive the signal including speech from the audio detection system, decompose the signal including speech into a sparse set of phonetic impulses, and access the database of keywords and convolve the sparse set of phonetic impulses with the ensemble of filters. The processor is further configured to identify keywords within the signal including speech based a result of the convolution and control operation the electronic system based on the keywords identified.

Computing the feature Maximum Mutual Information (fMMI) method requires multiplication of vectors with a huge matrix. The huge matrix is subdivided into block sub-matrices. The sub-matrices are quantized into different values and compressed by replacing the quantized element values with 1 or 2 bit indices. Fast multiplication with those compressed matrices with far fewer multiply/accumulate operations compared to standard matrix computation is enabled and additionally obviates a de-compression method for decompressing the sub-matrices before use.

According to an embodiment, a decoder includes a token operating unit, a node adder, and a connection detector. The token operating unit is configured to, every time a signal or a feature is input, propagate each of a plurality of tokens, which is an object assigned with a state of the of a path being searched, according to a digraph until a state or a transition assigned with a non-empty input symbol is reached. The node adder is configured to, in each instance of token propagating, add, in a lattice, a node corresponding to a state assigned to each of the plurality of tokens. The connection detector is configured to refer to the digraph and detect a node that is connected to a node added in an i-th instance in the lattice and that is added in an i+1-th instance in the lattice.

Systems and methods for spoken term detection are provided. A method for spoken term detection, comprises receiving phone level out-of-vocabulary (OOV) keyword queries, converting the phone level OOV keyword queries to words, generating a confusion network (CN) based keyword searching (KWS) index, and using the CN based KWS index for both in-vocabulary (IV) keyword queries and the OOV keyword queries.