TWI650749B - Voice processing device, voice recognition input system and voice recognition input method - Google Patents

Abstract

A speech recognition input system includes a speech processing device and a computer device. The voice processing device includes a microphone module, an audio transceiver module, a voice recognition module, a processing module, and a human-machine transmission interface. The microphone module outputs audio signals. The audio transceiver module sends audio signals to the speech recognition module. After the speech recognition module recognizes the audio signal, it correspondingly returns an identified document to the audio transceiver module. The processing module generates a message to be transmitted according to the identified document, and controls the human-machine transmission interface to output the message to be transmitted, where the message to be transmitted refers to a control instruction, text to be transmitted, or a combination thereof. The computer device includes a processor, and the processor receives the information to be transmitted, and executes the control instruction correspondingly, or displays the text to be transmitted or a combination thereof.

Description

Voice processing device, voice recognition input system and voice recognition input method

The invention relates to a recognition system, in particular to a speech processing device, a speech recognition input system and a speech recognition input method.

In the past, when a user operated a computer, they usually used a keyboard, mouse, or other peripheral device to complete the input of commands or text. However, with the development of speech recognition technology, many electronic devices (such as desktop computers, notebook computers, mobile phones, personal digital assistants or other electronic devices) also widely use speech as a method of text input.

In terms of application of speech recognition, most of them connect a computer with a microphone. The user can input audio through the microphone, and the audio can be recognized and converted into text by the speech recognition system for the computer to perform text input. However, the current microphones are generally connected to the computer through the UAC (USB Audio Class) device protocol, so it is easy to conflict with the original sound effects device of the computer system.

For example, assuming that a computer system has a built-in microphone, when the computer externally connects a microphone for voice input through the UAC device protocol, the computer system will display more than two microphone connections. At this time, the user must go to the system device option. Setting to select the microphone currently needed, otherwise the system will not be able to open the corresponding microphone or application correctly when inputting text by voice, resulting in music playback, real-time conversation may be voiced by the user, or an external microphone Occupied by other applications, preventing text input by voice.

On the other hand, after inputting voice through a microphone, it is recognized as text by a speech recognition system, and then the recognized text is input to a computer. However, most computers do not have a pre-loaded voice input method, so when you use voice input text, you need to install a voice input method or application on your computer, which is more inconvenient to use.

In view of this, the present invention provides a speech recognition input system including a speech processing device and a computer device. The voice processing device includes a microphone module, an audio transceiver module, a voice recognition module, a processing module, and a human-machine transmission interface (HID Interface). The microphone module outputs audio signals. The audio transceiver module is electrically connected to the microphone module and communicatively connected with the speech recognition module. The audio transceiver module receives and sends audio signals to the speech recognition module. After the speech recognition module recognizes the audio signal, it correspondingly returns an identified document to the audio transceiver module. The processing module is electrically connected to the audio transceiver module and the human-machine transmission interface. The processing module receives the identified document and generates a message to be transmitted according to the identified document. The message to be transmitted refers to a control command, a text to be transmitted, or a combination thereof , And the processing module controls the human-machine transmission interface to output the information to be transmitted. The computer device includes a processor, and the processor is electrically connected to the human-machine transmission interface to receive the information to be transmitted, and executes the control instruction correspondingly, or displays the text to be transmitted or a combination thereof.

In one embodiment, the present invention provides a voice processing device including a microphone module, an audio transceiver module, a processing module, and a HID Interface. The microphone module outputs audio signals. The audio transceiver module is electrically connected to the microphone module. The audio transceiver module transmits the audio signal after receiving the audio signal, and receives the identified file corresponding to the audio signal. The processing module is electrically connected to the audio transceiver module and receives the identified document. The processing module generates information to be transmitted according to the identified document. The information to be transmitted refers to a control instruction, text to be transmitted, or a combination thereof. The human-machine transmission interface is electrically connected to the processing module, and the processing module controls the human-machine transmission interface to output information to be transmitted.

In one embodiment, the present invention provides a voice recognition input method, including an audio output step: outputting an audio signal, and a voice recognition step: after the voice recognition module recognizes the audio signal and correspondingly outputs the identified document, the processing step: the processing module according to The identified file generates information to be transmitted, where the information to be transmitted refers to a control instruction, a text to be transmitted or a combination thereof, and an output step: the processing module controls a HID Interface to output the information to be transmitted.

In summary, the voice processing device according to the embodiment of the present invention is connected to a computer device through a human-machine transmission interface (HID Interface), so that the voice processing device and the computer device can transmit data through the HID communication protocol, compared with the previous microphone via UAC ( In terms of the USB Audio Class) device protocol and computer device connection, the voice processing device does not conflict with the built-in audio device of the computer device. The user does not need to select the audio device in the system device option. Therefore, the original computer device ’s Audio settings will not be changed, making it more convenient to use.

FIG. 1 is a system block diagram of the first embodiment of the speech recognition input system of the present invention. As shown in FIG. 1, the speech recognition input system 1 includes a speech processing device 10 and a computer device 20.

In some embodiments, the computer device 20 may be a personal computer, a notebook computer, a tablet computer, a smart phone, a navigation device, a car computer device, a personal digital assistant, a digital signage, or other devices capable of inputting text or instructions. Electronic device.

The voice processing device 10 can be connected to the computer device 20 to recognize audio (such as a user's voice or other sounds) and transmit commands to the computer device 20 to perform text input or perform actions. In some embodiments, the voice processing device 10 may be a computer peripheral device. For example, the computer peripheral device is a keyboard, a mouse, a stylus pen, or a speaker, so that the voice processing device 10 can control the computer manually. The device 20 performs text input or performs actions.

As shown in FIG. 1, in this embodiment, the voice processing device 10 includes a microphone module 11, an audio transceiver module 12, a processing module 14, and a human-machine transmission interface 15 (HID Interface). The microphone module 11 is used for receiving and converting an external speech (speech) and correspondingly outputting an audio signal S. For example, the microphone module 11 can capture a user's voice and correspondingly convert it into an audio signal S, such as a microphone module 11 A signal converter can be connected or built in to convert the user's voice (such as Fourier transform formula or other algorithms) into the audio signal S via the signal converter. In some embodiments, the microphone module 11 may be a dynamic microphone, a condenser microphone, an electret microphone, a micro-electro-mechanical microphone, or a directional microphone.

As shown in FIG. 1, the audio transceiver module 12 of the voice processing device 10 is electrically connected to the microphone module 11 to receive the audio signal S transmitted by the microphone module 11. In this embodiment, the audio transceiver module 12 is indirectly connected to the microphone module 11 through the processing module 14, but this is not limited, and the audio transceiver module 12 may also be directly connected to the microphone module 11. The audio transceiver module 12 can transmit an audio signal S to a voice recognition module 13 for voice recognition operations.

As shown in FIG. 1, in one embodiment, the voice processing device 10 may be connected to the voice recognition module 13 through the audio transceiver module 12. For example, the voice recognition module 13 may be a near-end voice recognition module, and the audio transceiver module 12 is a wired audio transceiver module, so as to be connected to the voice recognition module 13 through an electrical connection line. Or in the embodiment of FIG. 1, the speech recognition module 13 is a cloud speech recognition module and communicates with the Internet. For example, the speech recognition module 13 is a system built in the cloud, and the audio transceiver module 12 is Connected to the Internet in a wireless or wired manner to communicate with the voice recognition module 13. For example, the audio transceiver module 12 may be a wireless audio transceiver module, and the wireless audio transceiver module may be a wireless long-distance module (such as a 3G / 4G / 5G module, a radio module, or a wireless network module). Group) or a wireless short-range module (such as a WiFi module, a Bluetooth module, a ZigBee module), and wirelessly communicate with the voice recognition module 13 by being connected to the Internet in a wireless manner. Or, in another embodiment, the audio transceiver module 12 may also be connected to a wireless transmission device (such as a wireless sharer or a wireless base station) through an electrical connection line, so as to connect to the Internet via the wireless transmission device for voice recognition. The modules 13 communicate with each other.

In addition, in the embodiment of FIG. 1, the speech recognition module 13 is a cloud speech recognition module, so that the speech processing device 10 and the computer device 20 do not need to install additional speech recognition software, and the speech processing device 10 is connected to different When the computer device 20 is used, it is no longer necessary to perform voice training again, so that the voice processing device 10 can achieve the plug-and-play function and greatly improve the convenience in use.

In another embodiment, the speech recognition module 13 may also be built into the speech processing device 10. For example, FIG. 2 is a system block diagram of the second embodiment of the speech recognition input system of the present invention. The speech processing device 10 of the speech recognition input system 2 of this embodiment has a built-in speech recognition module 13 ', such as a speech recognition module. Group 13 'is hardware or firmware with computing capability (such as a digital signal processor or a programmable logic device, etc.), and the audio transceiver module 12 is a wired audio transceiver module, which is connected to the voice recognition module through an electrical connection line. Group 13 '.

As shown in FIG. 1 again, in this embodiment, after the speech recognition module 13 recognizes the audio signal S, it correspondingly returns an identified document T to the audio transceiver module 12. For example, the speech recognition module 13 may use statistical pattern recognition technology to perform speech recognition operations. For example, the speech recognition module 13 may have a built-in signal processing unit, an acoustic model, a pronunciation dictionary, a language model, and a decoder (the figure is omitted) (Illustrated), wherein the signal processing unit is used to extract features from the audio signal S for use by the acoustic model. The acoustic model may be modeled using, for example, a Hidden Markov Model (HMM) model. The language model models the target language. The pronunciation dictionary contains multiple vocabulary sets and their pronunciations to provide mapping between acoustic models and language models. The decoder searches for the text corresponding to the audio signal S according to the acoustic model, the language model and the pronunciation dictionary, so as to convert the audio signal S into a recognized document T (Text). The identified document T may include characters, symbols, punctuation marks, numbers, words, strings, or phrases composed of multiple strings, and so on.

As shown in FIG. 1, the processing module 14 is electrically connected to the audio transceiver module 12 and the human-machine transmission interface 15. The processing module 14 receives and generates an information to be transmitted I according to the identified file T, where the information to be transmitted I refers to A control instruction, a text to be transmitted, or a combination thereof. Specifically, the text to be transmitted may be in the form of plain text. For example, the text to be transmitted may include characters, symbols, punctuation marks, numbers, words, strings, or phrases composed of multiple strings. Text. The control instruction may be a signal form to control the computer device 20 to perform a specific action. For the computer device 20 as a personal computer, a notebook computer, or a smart phone, it may execute a file search, an opening / closing program, a disk reorganization, Specific actions such as volume control, shutdown, restart. If the computer device 20 is used as a navigation device, specific actions such as route planning, storage, and searching for parking spaces can be performed correspondingly. That is, the processing module 14 can determine whether it is a control instruction, a text to be transmitted, or a combination of the two according to the content of the recognized document T, which will be described in detail later.

In an embodiment, the processing module 14 of the speech processing device 10 may be specifically hardware capable of computing, such as a Central Processing Unit (CPU), or other programmable microprocessor (Microprocessor ), Digital Signal Processor (DSP), Programmable Controller, Application Specific Integrated Circuits (ASIC), Programmable Logic Device (PLD), or other similar devices .

As shown in FIG. 1 again, the computer device 20 includes a processor 21, and the processor 21 is electrically connected to the human-machine transmission interface 15 to receive the information I to be transmitted, and executes the control instruction correspondingly, or displays the text to be transmitted or a combination thereof. As a result, the voice processing device 10 according to the embodiment of the present invention is connected to the computer device 20 through a human-machine transmission interface 15 (HID Interface), so that the voice processing device 10 and the computer device 20 can transmit data through the HID communication protocol, compared to In the past, in the manner that the microphone was connected to the computer device 20 through a UAC (USB Audio Class) device protocol, the voice processing device 10 would not conflict with the audio devices (such as a microphone and a speaker) built in the computer device 20. In detail, assuming that the computer device 20 is a personal computer or a notebook computer, the system audio device will not show the addition of a new audio device, so it will not conflict with the built-in audio device of the computer device 20 and malfunction. The user does not need to select the audio device to be executed in the device option, thereby the audio setting of the original computer device 20 will not be changed, and the advantage of more convenience in use is achieved.

In order to explain the operation steps of the above embodiment more clearly, please refer to FIG. 5, where FIG. 5 is a flowchart of the steps of the first embodiment of the speech recognition input method of the present invention. The hardware structure mentioned below can be compared with the above speech recognition input. Revealed by System 1, first described here. In the process of voice recognition input of this embodiment, first, an audio output step S1 is performed: an audio signal S is output. Specifically, as shown in FIG. 1, the user can pass the voice processing device 10 through a human-machine The transmission interface 15 is connected to the computer device 20. When the user needs to control the computer device 20 to perform actions or input text, he can make a sound or speak a speech to the microphone module 11, and the microphone module 11 can capture and convert the sound Output for audio signal S.

After the audio output step S1, a voice recognition step S2 may be performed: the voice recognition module 13 recognizes the audio signal S and outputs a recognized file T correspondingly. For example, in the embodiments of FIGS. 1 and 2, the microphone module 11 can transmit the audio signal S to the audio transceiver module 12 to be transmitted to the cloud speech recognition module 13 via the audio transceiver module 12 (see FIG. 1). Or the speech recognition module 13 ′ (see FIG. 2) built in the speech processing device 10 recognizes the audio signal S and converts it into text to output the recognized document T.

After the speech recognition step S2, a processing step S3 may be performed: the processing module 14 generates a message I to be transmitted according to the recognized document T. For example, the processing module 14 may determine whether the information to be transmitted I is a control instruction, a text to be transmitted, or a combination of the two according to the content of the identified document T.

After processing step S3, an output step S4 may be performed: the processing module 14 controls a human-machine transmission interface 15 (HID Interface) to output information I to be transmitted. In this way, data can be transmitted between the voice processing device 10 and the computer device 20 via the HID communication protocol, so as to avoid conflicts with audio devices (such as microphones and speakers) built in the computer device 20, and the user does not need to go to the computer device Choose the audio device to be executed from the 20 device options, which is more convenient to use.

As shown in FIG. 5 again, after outputting step S4 in this embodiment, step S5 is further performed: the computer device 20 receives the information to be transmitted I, executes the control instruction correspondingly, or displays the text to be transmitted or a combination thereof. Specifically, when the processing module 14 outputs a control instruction, the computer device 20 performs a corresponding specific action, for example, a document search or a volume control operation. When the processing module 14 outputs the text to be transmitted, the computer device 20 correspondingly inputs the text to be transmitted and displays it on the screen. In an embodiment, the processing module 14 may also output control instructions and text to be transmitted at the same time, so as to drive the computer device 20 to perform actions and input text. For example, the computer device 20 may open a document and enter the text to be transmitted in the document, or the computer The device 20 starts a search engine and enters text to be transmitted for a search operation.

In some embodiments, the processing module 14 can determine whether the information to be transmitted I is a control instruction, a text to be transmitted, or a combination of the two according to the content of the identified document T in the following ways.

The processing module 14 can control the human-machine transmission interface 15 to output a text to be transmitted or output a control instruction according to whether the identified document T includes a trigger instruction. For example, FIG. 6 is a flowchart of steps in a second embodiment of a speech recognition input method according to the present invention. In this embodiment, the output step S4 may include three sub-steps. First, the sub-step S41: The processing module 14 may first determine whether the identified document T includes a trigger instruction. When the identified document T includes a trigger instruction, the sub-step is performed. S42: Control the human-machine transmission interface 15 to output the text to be transmitted. When the identified document T does not include a trigger instruction, perform sub-step S43: control the human-machine transmission interface 15 to output a control instruction. For example, please refer to FIG. 1. Assume that the word “speech input” is a condition that triggers the computer to input text. When the user speaks to the microphone module 11: “speech input, computer restarts”, speech recognition Module 13 can recognize the voice and convert it into a recognized document T in text form. Processing module 14 can generate the text to be transmitted according to the words of "voice input", where the text to be transmitted is the word "computer restart" String, because "voice input" is a trigger command that triggers computer input of text, the processing module 14 only controls the human-machine transmission interface 15 to output the word "computer restart" for input by the computer device 20 and is displayed on the screen Instead of the words "Speech, computer restart" on the computer screen. In another embodiment, assuming that the recognized document T does not include the above-mentioned "voice input", the processing module 14 controls the human-machine transmission interface 15 to output a control instruction corresponding to the content of "computer restart" to correspond to the computer The device 20 performs a restart operation. Of course, in other embodiments, the identified file T may also include trigger instructions and other control instructions (such as searching for files or volume control). The processing module 14 may simultaneously control the human-machine transmission interface 15 to output control instructions and to be transmitted. Text, this is not limited. In other embodiments, when the processing module 14 determines that the identified document T includes a trigger instruction, the control panel 15 is controlled to output a control instruction, and when the identified document T does not include a trigger instruction, it outputs Transfer text. Thus, in the embodiment of the present invention, the processing module 14 can determine whether the information to be transmitted I is a control instruction, text to be transmitted, or a combination of the two based on the content of the identified document T, and can prevent the computer device 20 from malfunctioning (for example, using When the user wants to control the computer device 20 to perform a specific action through voice, the computer device 20 enters text according to the voice content), so that the embodiment of the present invention can achieve a more accurate voice recognition input effect.

Alternatively, the speech processing device 10 of the speech recognition input system 3 may include a switch module 16, wherein the switch module 16 may be a button switch, a micro switch, a toggle switch, a membrane switch, a magnetic switch, etc., and the user can control the switch The module 16 corresponds to the output switching signal. The processing module 14 can control the human-machine transmission interface 15 to output the text to be transmitted or output a control instruction according to whether the switch signal is received. As shown in FIG. 7, it is a flowchart of steps in the third embodiment of the speech recognition input method of the present invention. In this embodiment, the output step S4 may include three sub-steps. First, sub-step S44: whether the processing module 14 receives a switch signal; if so, proceed to sub-step S45: control the human-machine transmission interface 15 to output the text to be transmitted. If not, proceed to sub-step S46: control the human-machine transmission interface 15 to output a control instruction. Specifically, please refer to FIG. 3, which is a system block diagram of the third embodiment of the speech recognition input system of the present invention. The user can manipulate the switch module 16 and speak a piece of text to the microphone module 11, and the speech recognition module 13 can recognize the above-mentioned speech and convert it into a recognized document T in text form. The processing module 14 can control according to the switch signal. The human-machine transmission interface 15 outputs the text to be transmitted (that is, the text). Conversely, when the user does not control the switch module 16 and only speaks the text to the microphone module 11, the processing module 14 outputs a control instruction according to the content of the text. In some embodiments, when the processing module 14 receives the switching signal, the human-machine transmission interface 15 may be controlled to output a control instruction; when the processing module 14 does not receive the switching signal, the human-machine transmission interface 15 may be controlled to output the standby instruction. Text is not limited.

As shown in FIG. 8, it is a flowchart of steps in the fourth embodiment of the speech recognition input method of the present invention. Compared with the embodiment of FIG. 5, the speech recognition step S2 ′ of this embodiment further includes: the semantic analysis unit 131 analyzes the recognized document T and outputs a command text, a non-command text or a combination thereof correspondingly. The processing step S3 'of this embodiment includes: the processing module 14 generates a control instruction according to the instruction text in the recognized document T, or generates a text to be transmitted according to a non-instruction text, or a combination thereof. For example, please refer to FIG. 4, which is a system block diagram of the fourth embodiment of the speech recognition input system of the present invention. In this embodiment, the speech recognition module 13 of the speech processing device 10 of the speech recognition input system 4 may include a semantic analysis unit 131 to correspondingly analyze the recognized document T. In some embodiments, the semantic analysis unit 131 may specifically be hardware capable of computing, such as a programmable microprocessor (Microprocessor), a digital signal processor (DSP), and the like.

In the past, semantic analysis technology refers to analyzing the content of a sentence, a string or a paragraph, and extracting the summary and the meaning from it. The use of semantic analysis does not limit the user's words, as long as the general language grammar is satisfied, it can be analyzed and discriminated. In some embodiments, the semantic analysis unit 131 may use singular value decomposition (SVD), non-negative matrix factorization (NMF), or neural network (NN) or other calculations. Method for semantic analysis. Specifically, the user's intention can be known by analyzing the semantics of the identified document T, such as when a string of the identified document T includes "search", "open", "close", "control", or " When adjusting verbs such as “adjustment”, you can use semantic analysis to determine that the user intends to manipulate the computer device 20 to perform actions, and then output the string as an instruction text, and the processing module 14 can generate the content of the instruction text correspondingly. Control instruction. When the word "input" is included in a string of the recognized document T, semantic analysis can be used to determine that the user has the intention to enter text, and then the string is output and represented as a non-command text. The processing module 14 can generate the text to be transmitted corresponding to the content of the non-command text.

As shown in FIG. 9, it is a flowchart of steps in a fifth embodiment of a speech recognition input method according to the present invention. Compared with the embodiment of FIG. 5, the processing step S3 '' in this embodiment further includes: compiling the text to be transmitted into a universal code format (Unicode). For example, in the embodiment of FIG. 3, the processing module 14 includes an encoding unit 141 (such as a text editor) to compile the text to be transmitted into a universal code format. Since the characters in the universal code format can be applied to various systems of the computer device 20, the computer device 20 can directly read the text to be transmitted and perform input operations, and does not need to additionally install other voice input methods, making voice input faster and more convenient.

As shown in FIG. 10, it is a flowchart of steps in a sixth embodiment of a speech recognition input method according to the present invention. Compared with the embodiment of FIG. 5, this embodiment further includes a translation step S6 after the speech recognition step S2: the translation unit 132 analyzes the audio signal S and outputs a translated text. The recognized document T includes a translation Text. Specifically, in the embodiment of FIG. 3, the speech recognition module 13 may include a translation unit 132 to decode the audio signal S and translate it into another language according to the needs of the user.

Although the technical content of the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art and making some changes and retouching without departing from the spirit of the present invention should be covered by the present invention. Therefore, the scope of protection of the present invention shall be determined by the scope of the appended patent application.

1 ~ 4‧‧‧speech recognition input system

10‧‧‧Voice processing device

11‧‧‧ Microphone Module

12‧‧‧Audio transceiver module

13, 13’‧‧‧ Voice recognition module

131‧‧‧ Semantic Analysis Unit

132‧‧‧ Translation Unit

14‧‧‧Processing Module

141‧‧‧coding unit

15‧‧‧ HMI

16‧‧‧Switch Module

20‧‧‧Computer device

21‧‧‧Processor

S‧‧‧Audio signal

T‧‧‧Identified documents

I‧‧‧ Information to be transmitted

S1‧‧‧Audio output steps

S2, S2’‧‧‧‧Speech recognition steps

S3, S3 ’, S3’’‧‧‧ processing steps

S4‧‧‧Output steps

S41 ~ S46‧‧‧‧Sub-step

S5‧‧‧ implementation steps

S6‧‧‧Translation steps

[FIG. 1] A system block diagram of the first embodiment of the speech recognition input system of the present invention. [FIG. 2] It is a system block diagram of the second embodiment of the speech recognition input system of the present invention. [FIG. 3] A system block diagram of the third embodiment of the speech recognition input system of the present invention. [FIG. 4] A system block diagram of the fourth embodiment of the speech recognition input system of the present invention. [FIG. 5] It is a flowchart of steps in the first embodiment of the speech recognition input method of the present invention. [FIG. 6] It is a flowchart of the steps of the second embodiment of the speech recognition input method of the present invention. [FIG. 7] It is a flowchart of the steps of the third embodiment of the speech recognition input method of the present invention. [FIG. 8] It is a flowchart of the steps of the fourth embodiment of the speech recognition input method of the present invention. [FIG. 9] It is a flowchart of steps in the fifth embodiment of the speech recognition input method of the present invention. [FIG. 10] It is a flowchart of steps in the sixth embodiment of the speech recognition input method of the present invention.

Claims (27)

A voice recognition input system includes: a voice processing device including a microphone module, an audio transceiver module, a voice recognition module, a processing module, and a HID Interface. The microphone module outputs An audio signal, the audio transceiver module is electrically connected to the microphone module, and is communicatively connected with the voice recognition module, the audio transceiver module receives and sends the audio signal to the voice recognition module, and the voice recognition module recognizes The audio signal correspondingly returns an identified document to the audio transceiver module, the processing module electrically connects the audio transceiver module and the human-machine transmission interface, and the processing module receives the identified document, and according to the The identification file generates information to be transmitted, the information to be transmitted refers to a control instruction, a text to be transmitted, or a combination thereof, and the processing module controls the human-machine transmission interface to output the information to be transmitted; and a computer device including a A processor, which is electrically connected to the human-machine transmission interface to receive the information to be transmitted, and executes the control instruction correspondingly, or displays the to-be-transmitted information Text, or combinations thereof. The speech recognition input system according to claim 1, wherein the audio transceiver module is a wireless audio transceiver module. The speech recognition input system according to claim 1, wherein the speech recognition module is a cloud speech recognition module and the communication is connected to the Internet, and the audio transceiver module is connected to the Internet in a wireless or wired manner to Communicate with the speech recognition module. The speech recognition input system according to claim 1, wherein the speech recognition module includes a translation unit that analyzes the audio signal and outputs a translated text, and the recognized document includes the translated text. The speech recognition input system according to claim 1, wherein the speech recognition module includes a semantic analysis unit, the semantic analysis unit analyzes the recognized document and outputs a command text, a non-command text, or a combination thereof, and the processing The module generates the control instruction according to the command text of the recognized document, or generates the text to be transmitted according to the non-command text of the identified document, or a combination thereof. The speech recognition input system according to claim 1, wherein the processing module controls the output of the text to be transmitted or the control instruction according to whether the recognized document includes a trigger instruction. The speech recognition input system according to claim 1, wherein the speech processing device includes a switch module, and the switch module is connected to the processing module and can selectively output a switching signal. The switch signal controls outputting the text to be transmitted or outputting the control instruction. The speech recognition input system according to claim 1, wherein the processing module includes an encoding unit, and the encoding unit compiles the text to be transmitted into a universal code format (Unicode). The speech recognition input system according to claim 1, wherein the speech processing device is a computer peripheral device, and the computer peripheral device is a keyboard, a mouse, a stylus, or a speaker. A voice processing device includes: a microphone module that outputs an audio signal; an audio transceiver module that is electrically connected to the microphone module, the audio transceiver module transmits the audio signal after receiving the audio signal, and receives the audio signal via An identified document generated after speech recognition; a processing module electrically connected to the audio transceiver module and receiving the identified document; the processing module generates a to-be-transmitted information according to the identified document, the to-be-transmitted information It refers to a control instruction, a text to be transmitted, or a combination thereof; and a human-machine transmission interface (HID Interface), which is electrically connected to the processing module, and the processing module controls the human-machine transmission interface to output the information to be transmitted. The voice processing device according to claim 10, wherein the audio transceiver module is a wired audio transceiver module. The voice processing device according to claim 10, further comprising a voice recognition module, which is communicatively connected to the audio transceiver module, and the voice recognition module receives the audio signal and recognizes it and outputs the recognized document correspondingly. The speech processing device according to claim 12, wherein the speech recognition module is a cloud speech recognition module and is communicatively connected to the Internet, and the audio transceiver module is wirelessly or wiredly connected to the Internet to communicate with the Internet. The speech recognition modules communicate with each other. The speech processing device according to claim 12, wherein the speech recognition module includes a translation unit, the analysis unit analyzes the audio signal and outputs a translated text, and the recognized document includes the translated text. The speech processing device according to claim 12, wherein the speech recognition module includes a semantic analysis unit that analyzes the recognized document and correspondingly outputs a command text, a non-command text, or a combination thereof. The processing module The group is to generate the control instruction according to the instruction text of the identified document, or to generate the text to be transmitted, or a combination thereof, according to the non-instruction text of the identified document. The speech processing device according to claim 10, wherein the processing module includes an encoding unit, and the encoding unit compiles the text to be transmitted into a universal code format (Unicode). The speech processing device according to claim 10, wherein the processing module controls the output of the text to be transmitted when determining that the recognized document includes a trigger instruction, and the processing module determines that the recognized document does not include the When a command is triggered, the control outputs the control command. The voice processing device according to claim 10, further comprising a switch module. The switch module is electrically connected to the processing module and can selectively output a switching signal. When the processing module receives the switching signal, it controls When the text to be transmitted is output, the control module outputs the control instruction when the processing module does not receive the switch signal. The voice processing device according to claim 10, wherein the voice processing device is a computer peripheral device, and the computer peripheral device is a keyboard, a mouse, a stylus, or a speaker. A voice recognition input method includes the following steps: an audio output step: outputting an audio signal; a voice recognition step: a voice recognition module recognizes the audio signal and correspondingly outputs an identified document; a processing step: a processing module according to the The identified file generates a message to be transmitted, wherein the message to be transmitted refers to a control instruction, a text to be transmitted, or a combination thereof; and an output step: the processing module controls a HID Interface to output the information to be transmitted . The speech recognition input method according to claim 20, wherein the speech recognition module in the speech recognition step is a cloud speech recognition module and the communication is connected to the Internet. The speech recognition input method according to claim 20, wherein the speech recognition step includes: a semantic analysis unit analyzes the recognized document and correspondingly outputs a command text, a non-command text or a combination thereof, and the processing step includes the processing The module generates the control instruction according to the instruction text, or generates the text to be transmitted according to the non-instruction text, or a combination thereof. The speech recognition input method according to claim 20, wherein the speech recognition step further includes a translation step: a translation unit analyzes the audio signal and outputs a translated text, and the recognized document includes the translated text. The speech recognition input method according to claim 20, wherein the processing step includes: compiling the text to be transmitted into a universal code format (Unicode). The speech recognition input method according to claim 20, wherein the output step includes: the processing module controls outputting the text to be transmitted or outputting the control instruction according to whether the recognized document includes a trigger instruction. The speech recognition input method according to claim 20, wherein the outputting step includes: the processing module controls outputting the text to be transmitted or outputting the control instruction according to whether a switching signal is received from a switching module. The speech recognition input method according to claim 20, further comprising an execution step: a computer device receives the information to be transmitted, and executes the control instruction correspondingly, or displays the text to be transmitted or a combination thereof.