Speech recognition of partial proper names by natural language processing

• US 9,589,563 B2
• Filed: 06/02/2015
• Issued: 03/07/2017
• Est. Priority Date: 06/02/2014
• Status: Active Grant

• Current Assignee: Robert Bosch GmbH
• Original Assignee: Robert Bosch GmbH
• Inventors: Zhao, Lin, Feng, Zhe, Xu, Kui, Weng, Fuliang
• Primary Examiner(s): Desir, Pierre-Louis
• Assistant Examiner(s): SHIN, SEONG-AH A
• Application Number:
  US14/728,118
• Publication Number:
  US 20150348543A1
• Time in Patent Office: 644 Days
• Field of Search: 704/9, 704/256.5, 704/257, 704/4, 704/201, 704/231, 704/235, 704/240, 704/251, 704/254
• US Class Current: 1/1
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules

Abstract

A method for speech recognition of partial proper names is described which includes natural language processing (NLP), partial name candidate generation, speech recognition and post processing. Natural language processing techniques including shallow and deep parsing are applied to long proper names to identify syntactic units (for example, noun phrases). The syntactic units form a basis for generating a candidate list of partial names for each original full name. A partial name is part of the original name, with some words omitted, or word order changed, or even word substitution. After candidate partial names are generated, their phonetic transcriptions are incorporated into a model for a speech recognizer to recognize the partial names in a speech recognition system.

Claims

1. A method of speech recognition for partial proper names comprising:

receiving, with a processor, a corpus including a plurality of complex proper names;

performing, with the processor, natural language processing to generate a syntactic structure corresponding to a plurality of words in one of the plurality of complex proper names in the corpus;

generating, with the processor, a plurality of candidate partial names corresponding to the one complex proper name using a machine learning process with reference to the syntactic structure corresponding to the one complex proper name and the plurality of words in the one complex proper name, wherein the one complex proper name is divided into categories of syntactic units and each of the plurality of candidate partial names comprises a subset of words from the plurality of words contained within the one complex proper name and the subset of words classified into a specific syntactic category;

selecting, with the processor, only a portion of the plurality of candidate partial names based on at least one syntactic structural identifier of at least one phrase in each candidate partial name to provide a modified list of candidate partial names;

producing, with the processor, a phonetic transcription of each partial name in the modified list of candidate partial names;

incorporating, with the processor, the modified list of candidate partial names based on the phonetic transcription into a model for a speech recognizer to recognize partial names in a speech recognition process; and

storing, with the processor, the model for the speech recognizer in a memory;

receiving, with the processor, speech input from a user comprising a candidate partial name from the modified candidate partial names;

performing, with the processor, speech recognition, using the model, on the received speech; and

identifying, with the processor, the one complex proper name based on the recognized speech and performing an action on a user device based on the identified one complex proper name.

2. The method of claim 1, the natural language processing further comprising:

generating with the processor the syntactic structure using Part-of-Speech (POS) tagging.

3. The method of claim 1, the natural language processing further comprising:

generating with the processor the syntactic structure using shallow chunking.

4. The method of claim 1, the natural language processing further comprising:

generating with the processor the syntactic structure using syntactic parsing.

5. The method of claim 1, the machine learning process further comprising:

generating with the processor the plurality of partial proper name candidates using Conditional Maximum Entropy (CME).

6. The method of claim 1, the machine learning process further comprising:

generating with the processor the plurality of partial proper name candidates using a Conditional Random Field (CRF).

7. The method of claim 1, the machine learning process further comprising:

generating with the processor the plurality of partial proper name candidates using a Hidden Markov Model (HMM).

8. The method of claim 1, the machine learning process further comprising:

generating with the processor the plurality of partial proper name candidates using a neural network.

9. A speech recognition system comprising:

a memory configured to store;

a corpus including a plurality of complex proper names; and

a model for a speech recognizer; and

a processor operatively connected to the memory, the processor being configured to;

perform natural language processing to generate syntactic structure corresponding to a plurality of words in one of the plurality of complex proper names in the corpus;

generate a plurality of candidate partial names corresponding to the one complex proper name using a machine learning process with reference to the syntactic structure corresponding to the one complex proper name and the plurality of words in the one complex proper name, wherein the one complex proper name is divided into categories of syntactic units and each of the plurality of candidate partial names comprises a subset of words from the plurality of words contained within the one complex proper name and the subset of words classified into a specific syntactic category;

select only a portion of the plurality of candidate partial names based on at least one syntactic structural identifier of at least one phrase in each candidate partial name to provide a modified list of candidate partial names;

incorporate the modified list of candidate partial names based on the phonetic transcription into the model for the speech recognizer to recognize partial names in a speech recognition process;

store the model for the speech recognizer in the memory;

receive speech input from a user comprising a candidate partial name from the modified candidate partial names;

perform speech recognition, using the model, on the received speech; and

identify the one complex proper name based on the recognized speech and performing an action on a user device based on the identified one complex proper name.

10. The system of claim 9, the processor being further configured to:

perform the natural language processing to generate the syntactic structure using Part-of-Speech (POS) tagging.

11. The system of claim 9, the processor being further configured to:

perform the natural language processing to generate the syntactic structure using shallow chunking.

12. The system of claim 9, the processor being further configured to:

perform the natural language processing to generate the syntactic structure using syntactic parsing.

13. The system of claim 9, the processor being further configured to:

generate the plurality of partial proper name candidates using Conditional Maximum Entropy (CME).

14. The system of claim 9, the processor being further configured to:

generate the plurality of partial proper name candidates using a Conditional Random Field (CRF).

15. The system of claim 9, the processor being further configured to:

generate the plurality of partial proper name candidates using a Hidden Markov Model (HMM).

16. The system of claim 9, the processor being further configured to:

generate the plurality of partial proper name candidates using a neural network.