# US 9,437,217 B2 · Pre-processing apparatus and method for speech recognition

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

- Number: US 9,437,217 B2
- Document ID: US9437217B2
- Status: Granted
- Kind: B2
- Application number: 14/484,084
- Filed: 2014-09-11
- Published: 2015-05-21
- Granted: 2016-09-06
- Earliest claimed: 2013-11-15
- Assignees: Hyundai Mobis Co., Ltd.
- Applicants: Hyundai Mobis Co., Ltd.
- Inventors: Min Ho Kwon
- Web page: https://patentgrep.com/patent/US9437217B2

## CPC

- G10L 25/78: Detection of presence or absence of voice signals (switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10)
- G10L 15/00: Speech recognition (G10L17/00 takes precedence)
- G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L 21/045: using thinning out or insertion of a waveform
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

A pre-processing apparatus for speech recognition may include: a trailing silence period detection unit configured to detect the length of a trailing silence period contained in a speech signal; a reference trailing silence period storage unit configured to store the length of a reference trailing silence period; and a trailing silence period adjusting unit configured to adjust the length of the trailing silence period contained in the speech signal based on the length of the reference trailing silence period.

## Claims

### Claim 1

A speech recognition system for a vehicle, the system comprising:
a speech recognition module configured to analyze voice data for recognition of a voice command, wherein the voice recognition module's voice recognition accuracy varies depending on length of a leading silence contained in the voice data, wherein the leading silence is defined as a period prior to the first signal reaching a predetermined intensity in the voice data;
a voice capturing module configured to capture a voice command subsequent to a voice command prompt sound generated by the speech recognition system; and
a voice data module configured to analyze captured data of the voice command and to determine whether to adjust a leading silence in the captured data to have a predetermined length to improve speech recognition accuracy of the speech recognition module.

### Claim 2 (depends on 1)

The speech recognition system of claim 1, wherein the voice data module is configured to lengthen the leading silence when the leading silence is shorter than the predetermined length in the captured data.

### Claim 3 (depends on 2)

The speech recognition system of claim 2, wherein the voice data module is configured to shorten the leading silence when the leading silence is longer than the predetermined length in the captured data.

### Claim 4

A speech recognition method for a vehicle, the method comprising:
generating a voice command prompt sound while the vehicle is running;
subsequently capturing a voice command to provide a voice data;
analyzing the voice data and determining whether to adjust a leading silence in the voice data to have a predetermined length for improve speech recognition accuracy of a speech recognition module, wherein the leading silence is defined as a period prior to the first signal reaching a predetermined intensity in the voice data;
modifying the voice data to include a leading silence of the predetermined length; and
analyzing modified voice data to recognize of the voice command using the speech recognition module.

### Claim 5 (depends on 4)

The method of claim 4, wherein modifying the voice data comprises lengthening the leading silence when the leading silence is shorter than the predetermined length in the voice data.

### Claim 6 (depends on 4)

The pre processing method of claim 4, wherein modifying the voice data comprises shortening the leading silence when the leading silence is longer than the predetermined length in the voice data.