# US 2019/0172447 A1 · Speech recognition circuit using parallel processors

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0172447 A1
- Document ID: US20190172447A1
- Status: Published application
- Kind: A1
- Application number: 16/266,265
- Filed: 2019-02-04
- Published: 2019-06-06
- Earliest claimed: 2002-02-04
- Assignees: Zentian Limited
- Applicants: Zentian Limited
- Inventors: Mark Catchpole
- Web page: https://patentgrep.com/patent/US20190172447A1

## CPC

- G10L 15/187: Phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams
- G10L 15/05: Word boundary detection
- G10L 15/34: Adaptation of a single recogniser for parallel processing, e.g. by use of multiple processors or cloud computing
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## 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. A controller controls the lexical tree processors to process lexical trees identified in the results memory by performing parallel processing on a plurality of said lexical tree data structures.

## Claims

### Claim 1

A speech recognition circuit comprising:
one or more clusters of processors, each said cluster comprising:
one or more processors; and
an acoustic model memory storing acoustic model data;
wherein each said processor is configured to compute a probability using the said acoustic model data in said acoustic model memory;

### Claim 2 (depends on 1)

The speech recognition circuit of claim 1 where said probability is an input to an evaluation of a state transition of a model of states.

### Claim 3 (depends on 2)

The speech recognition circuit of claim 2 where said model is a Hidden Markov Model.

### Claim 4 (depends on 2)

The speech recognition circuit of claim 2 where said probability is computed from a Gaussian mixture model and one or more feature vectors.

### Claim 5 (depends on 1)

The speech recognition circuit of claim 1 where the circuit further comprising:
a buffer for storing one or more feature vectors coupled to said processors.

### Claim 6 (depends on 5)

The speech recognition circuit of claim 5 further comprising:
a search controller coupled to said processors and capable of controlling said processors to initiate speech recognition processing in all said processors or in a subset of said processors.

### Claim 7 (depends on 1)

The speech recognition circuit of claim 1 wherein the said acoustic model memory in each said cluster of processors is a distinct and separate memory from the said acoustic model memory is each other ones of said clusters of processors.

### Claim 8 (depends on 1)

The speech recognition circuit of claim 1 where said acoustic model memory is coupled to all of said processors in all of said clusters of processors.
