# US 2021/0343270 A1 · Speech translation method and translation apparatus

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

- Number: US 2021/0343270 A1
- Document ID: US20210343270A1
- Status: Published application
- Kind: A1
- Application number: 16/470,560
- Filed: 2019-04-02
- Published: 2021-11-04
- Earliest claimed: 2018-09-19
- Assignees: Langogo Technology Co., Ltd
- Applicants: Langogo Technology Co., Ltd
- Inventors: YAN ZHANG; Tao Xiong
- Web page: https://patentgrep.com/patent/US20210343270A1

## CPC

- G06F 40/58: Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
- G10L 15/005: {Language recognition}
- G10L 15/08: Speech classification or search
- G10L 17/00: Speaker identification or verification techniques
- Y02D 10/00: Energy efficient computing, e.g. low power processors, power management or thermal management
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- Y02D: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE

## Abstract

There are a speech translation method and a translation apparatus. The method includes: collecting a sound in response to a translation task being triggered, and detecting whether a user starts speaking based on the collected sound; entering a voice recognition state in response to detecting the user having started speaking, extracting a user voice from the collected sound, determining a source language used by the user based on the extracted user voice, and determining a target language associated with the source language based on a preset language pair; exiting the voice recognition state in response to detecting the user having stopped speaking for more than a preset delay duration, and converting the user voice extracted in the voice recognition state into a target voice of the target language; a playing the target voice, and returning to the step of detecting whether the user starts speaking until the translation task ends.

## Claims

### Claim 1

A speech translation method for a speech translation apparatus, wherein the translation apparatus comprising a processor, a sound collecting device electrically coupled to the processor, and a sound playback device electrically coupled to the processor; wherein the method comprises:
collecting a sound in an environment through the sound collecting device in response to a translation task being triggered, and detecting whether a user starts speaking based on the collected sound through the processor;
entering a voice recognition state in response to detecting the user having started speaking, extracting a user voice from the collected sound through the processor, determining a source language used by the user based on the extracted user voice, and determining a target language associated with the source language based on a preset language pair;
exiting the voice recognition state in response to detecting the user having stopped speaking for more than a preset delay duration, and converting the user voice extracted in the voice recognition state into a target voice of the target language through the processor; and
playing the target voice through the sound playback device, and returning to the step of detecting whether the user starts speaking based on the collected sound through the processor until the translation task ends.

### Claim 2 (depends on 1)

The method of claim 1, wherein before the step of entering the voice recognition state in response to detecting the user having started speaking further comprises:
detecting whether a noise in the environment is greater than a preset noise based on the collected sound through the processor, and outputting prompt information for prompting the user the environment being unsuitable for translations if the noise is greater than the preset noise.

### Claim 3 (depends on 1)

The method of claim 1, wherein the method further comprises:
setting at least two languages specified by a language specifying operation as the language pair through the processor, in response to the language specifying operation of the user.

### Claim 4 (depends on 1)

The method of claim 1, wherein the translation apparatus further comprises a display screen electrically coupled to the processor, after the steps of entering the voice recognition state in response to detecting the user having started speaking and extracting the user voice from the collected sound through the processor further comprises:
converting the extracted user voice into a corresponding first text, and displaying the first text on the display screen;
the steps of exiting the voice recognition state in response to detecting the user having stopped speaking for more than the preset delay duration and converting the user voice extracted in the voice recognition state into the target voice of the target language through the processor specifically comprises:
exiting the voice recognition state in response to detecting the user having stopped speaking for more than the preset delay duration, translating the first text into a second text of the target language through the processor, and displaying the second text on the display screen; and
converting the second text into the target voice through a speech synthesis system.

### Claim 5 (depends on 1)

The method of claim 1, wherein before the step of exiting the voice recognition state in response to detecting the user having stopped speaking for more than the preset delay duration further comprises:
exiting the voice recognition state in response to a translation instruction being triggered; and
adjusting the preset delay duration based on a time difference between a time of having detected the user having stopped speaking and a time of the translation instruction being triggered.

### Claim 6 (depends on 5)

The method of claim 5, wherein the translation apparatus further comprises a motion sensor electrically coupled to the processor, the method further comprises:
triggering the translation instruction in the voice recognition state, when a motion amplitude of the translation apparatus detected through the motion sensors is greater than a preset amplitude or the translation apparatus is collided.

### Claim 7 (depends on 5)

The method of claim 5, wherein the translation apparatus further comprises a storage electrically coupled to the processor, the step of determining the source language used by the user based on the extracted user voice further comprises:
extracting a voiceprint feature of the user in the user voice through the processor, and determining whether identifier information of a language corresponding to the voiceprint feature is stored in the storage;
determining a language corresponding to the identifier information as the source language, if the identifier information is stored in the storage; and
extracting a pronunciation feature of the user in the user voice, determining the source language based on the pronunciation feature, and storing a correspondence between the voiceprint feature of the user and the identifier information of the source language in the storage, if the identifier information is not stored in the storage.

### Claim 8 (depends on 7)

The method of claim 7, wherein the step of adjusting the preset delay duration based on the time difference between the time of having detected the user having stopped speaking and the time of the translation instruction being triggered specifically comprises:
determining whether the preset delay duration corresponding to the voiceprint feature of the user having stopped speaking is stored in the storage;
adjusting the corresponding preset delay duration based on the time difference between the time of having detected the user having stopped speaking and the time of the translation instruction being triggered, if the corresponding preset delay duration is stored in the storage; and
setting the time difference as the corresponding preset delay duration, if the corresponding preset delay duration is not stored in the storage.

### Claim 9

A translation apparatus, wherein the apparatus comprises:
an end point detecting module configured to collect a sound in an environment through the sound collecting device in response to a translation task being triggered, and detect whether a user starts speaking based on the collected sound;
a recognition module configured to enter a voice recognition state in response to detecting the user having started speaking, extract a user voice from the collected sound, determine a source language used by the user based on the extracted user voice, and determining a target language associated with the source language based on a preset language pair;
a tail point detecting module configured to detect whether the user has stopped speaking for more than a preset delay duration, and exit the voice recognition state in response to detecting the user having stopped speaking for more than the preset delay duration;
a translation and voice synthesizing module configured to convert the user voice extracted in the voice recognition state into a target voice of the target language through the processor; and
a playback module configured to play the target voice through the sound playback device, and trigger the end point detecting module to execute the step of detecting whether the user starts speaking based on the collected sound.

### Claim 10

A translation apparatus, wherein the apparatus comprises a sound collecting device, a sound playback device, a storage, a processor, and a computer program stored in the storage and executable on the processor;
wherein, the sound collecting device, the sound playback device, and the storage are electrically coupled to the processor;
when the processor executes the computer program, the following steps are executed:
collecting a sound in an environment through the sound collecting device in response to a translation task being triggered, and detecting whether a user starts speaking based on the collected sound;
entering a voice recognition state in response to detecting the user having started speaking, extracting a user voice from the collected sound, determining a source language used by the user based on the extracted user voice, and determining a target language associated with the source language based on a preset language pair;
exiting the voice recognition state in response to detecting the user having stopped speaking for more than a preset delay duration, and converting the user voice extracted in the voice recognition state into a target voice of the target language; and
playing the target voice through the sound playback device, and returning to the step of detecting whether the user starts speaking based on the collected sound until the translation task ends.
