WO2020006676A1

WO2020006676A1 - Blind input method using touch soft keyboard, computing apparatus, and storage medium

Info

Publication number: WO2020006676A1
Application number: PCT/CN2018/094203
Authority: WO
Inventors: 喻纯; 王丰; 石伟男; 范舒翼; 王童; 史元春
Original assignee: 清华大学
Priority date: 2018-07-03
Filing date: 2018-07-03
Publication date: 2020-01-09

Abstract

Provided are an input method for implementing input using a touch soft keyboard, a computing apparatus, and a computer readable medium. The input method comprises determining the coordinate position of a finger of a user on a screen in response to a touch input signal from a touch sensor; on the basis of the touch position, calculating the input probability of each character; on the basis of the input probability value of each character, determining the optimum character, and playing a sound of the optimum character. Furthermore, determining the optimum character comprises: on the basis of the input probability value of each character, determining the character having the highest probability value to be a candidate optimum character; on the basis of the touch press position and the determined candidate optimum character, attempting to update the keyboard layout, such that the candidate optimum character is in the touch press position and the updated keyboard layout meet a predetermined standard; if the keyboard layout update is successful, then determining that the candidate optimum character is the optimum character; and, if not, then using a default standard keyboard layout, and determining that the optimum character is the button in the touch press position in said layout.

Description

Blind input method using touch soft keyboard, computing device and storage medium

Technical field

The present invention generally relates to information input technology of a computing device, and particularly to a blind input technology using a soft keyboard on a touch screen device, a computing device, and a computer storage medium.

Background technique

Text input on a touch screen device is one of the most basic functions of a smart device.

There are existing technologies, such as CN102915122A, that use language models to achieve predictions of inputs. This prior art defaults the correctness of the typing operation and feedbacks through vision, which is not suitable for blind input applications.

There are also technologies that use direct screen reading to enable blind people to input on touch-screen devices, such as the patent CN103064529A. This prior art word selection operation is the same as typing operation on the screen to search with fingers, increasing the possibility of misoperation. , The practicability of blind input is low; and the intent of the blind cannot be predicted, error correction cannot be performed, the accuracy of the input is reduced, and the time-consuming to retrieve the keys is increased.

Summary of the invention

In view of the foregoing, the present invention has been made.

According to an aspect of the present invention, there is provided an input method using a touch-type soft keyboard in a computing device, comprising: determining a coordinate position of a user's finger on a screen in response to a touch input signal from a touch sensor; and according to the touch position , Calculate the input probability of each character; determine the optimal character according to the input probability value of each character, and play the voice of the optimal character.

Optionally, determining the optimal character includes: determining, according to the input probability value of each character, the character with the highest probability value as the candidate optimal character; and trying to update the keyboard layout according to the touch-down position and the determined candidate optimal character, So that the candidate optimal character is at the touch-press position and the updated keyboard layout meets a predetermined standard; if the updated keyboard layout is successful, the candidate optimal character is determined to be the optimal character; otherwise, the default standard keyboard layout is used to determine The optimal character is the key at the touched position in this layout. .

Optionally, the input method further includes: in response to the user continuing to move to access other key characters without lifting the finger, playing the voice of the key character corresponding to the finger position.

Optionally, the input method further includes: after playing the voice of the key character, in response to detecting a confirmation signal originating from the user, completing the input of the confirmed character.

Optionally, the confirmation signal includes: a signal triggered by the user lifting a finger or a signal triggered by a double-click operation after lifting the finger.

Optionally, determining the input probability of each character includes: calculating the input probability of each character based on the character that has been input and the position where the user touched and pressed, using a drop point model and a language model.

Optionally, before calculating the input probability of each character according to the touch position, the method further includes: determining whether the distance between the current touch position and the previous confirmed input touch position is less than a certain threshold, and if so, directly using the previous Confirm the keyboard layout at the time of input, and do not update the keyboard layout; if not, restore the keyboard layout to the default standard keyboard layout, and then do probability calculations and determine the optimal characters, and update the keyboard layout as needed based on the results.

Optionally, the keyboard layout is an invisible keyboard response area setting capable of achieving the same effect.

Optionally, the landing point model is a normal distribution.

Optionally, the updating the keyboard layout includes moving and / or compressing the keyboard.

Optionally, the moving and / or compressing the keyboard includes: performing branch or overall movement and compression; compressing the keyboard in a linear or exponential manner with the touch-pressed position as the center; the movement must not exceed a predefined effective range; the compression must not Make any key size smaller than the size threshold.

Optionally, the input method further includes determining whether a gesture input has occurred, the determining whether the gesture input has occurred includes: detecting a contact trajectory within a full screen range instead of being limited to an area occupied by a soft keyboard; using a displacement of the contact trajectory- The time relationship determines whether a gesture input has occurred and determines the type of gesture; based on the determined gesture type, a corresponding input operation is determined.

Optionally, gesture types and corresponding input operations include: single-finger swipe left: delete the last typed character; two-finger swipe left: clear all typed characters; single-finger swipe up: switch the highlighted candidate to the previous candidate Words: Single-finger swipe: Switch the highlighted candidate to the next candidate; Single-finger swipe right: Confirm input of the current candidate

Optionally, the input method further includes: providing audio feedback for touch operations, including: feedback on key presses; feedback on confirming input; and response to gesture operations.

According to another aspect of the present invention, there is also provided a computing device including: a touch screen with a touch sensor capable of detecting a touch input; a controller and a memory, wherein the computer-executable instructions are stored on the memory, and the touch-type soft keyboard is activated In the case of: when the computer-executable instructions are executed by a controller, they are operable to perform the aforementioned input method.

According to another aspect of the present invention, there is provided a computer-readable storage medium having computer-executable instructions stored thereon, and in a case where a touch-type soft keyboard of a computing device has been activated: When executed, it is operable to perform the aforementioned input method.

According to the input method, the computing device and the computer-readable medium of the embodiment of the present invention, the feedback of the input method is completely audioized, and the prediction algorithm is adjusted. The soft keyboard zoom and translation can be implemented according to the characteristics of blind input, which can effectively improve Input accuracy and speed of blind people on touch screen devices. The techniques used in this example are based on conditional probability formulas, Bayesian formulas, text-to-speech synthesis, and so on.

BRIEF DESCRIPTION OF THE DRAWINGS

These and / or other aspects and advantages of the present invention will become clearer and easier to understand from the following detailed description of the embodiments of the present invention in conjunction with the accompanying drawings, in which:

FIG. 1 illustrates an overall flowchart of an input method 100 for inputting by using a touch-type soft keyboard in a computing device according to an embodiment of the present invention.

The table shown in Figure 2 gives a schematic example of a language model.

The table shown in Figure 3 gives a schematic example of the fall point model.

FIG. 4 shows an overall flow of a text input method 200 according to another embodiment of the present invention.

Figures 5a and 5b show examples of soft keyboard movement and zooming processes in actual use.

FIG. 6 shows a schematic example of illegal movement of a soft keyboard in actual use.

Fig. 7a shows the keyboard state before the user presses a key in an input example, Fig. 7b shows the layout of the keyboard after the user first presses a key when entering a letter, and Fig. 7c shows an error that occurs when "o" is entered situation. Figure 7d uses the left swipe gesture to delete, at this time the voice feedback: "o has been deleted".

detailed description

In order to enable those skilled in the art to better understand the present invention, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

Before introducing, explain the meaning of related terms in this article.

Touch-press: refers to the action from the finger never touching the touch screen to touching the touch screen.

Movement: Refers to the movement of the finger while changing the position while keeping touching the touch screen.

Lift: refers to the movement of a finger from touching the touch screen to not touching the touch screen.

Specific embodiments of the present invention will be described below with reference to the drawings.

The input method of this embodiment is particularly suitable for inputting text. There is no limitation on the type of input text. For example, the input method may be English character input or Chinese character input.

As shown in FIG. 1, in step S110, in response to a touch input signal from a touch sensor, a coordinate position of a user's finger on a screen is determined.

In step S120, the input probability of each character is calculated based on the touched position.

In one example, the input probability of each character currently input by the user is calculated based on the character string that has been input and the position where the user touches and presses, and uses a drop point model and a language model.

In one example, the language model is a Chinese / English dictionary consisting of vocabulary and frequency pairs. The table shown in Figure 2 gives a schematic example of a language model.

In one example, the landing point model is a combined list of expected input keys and actual input key distributions on a touch-screen device when the user has no visual feedback. This model can be obtained through testing and statistics of user input behavior. The table shown in Figure 3 gives a schematic example of the fall point model.

In one example, by combining the calculation of the language model and the point model to correct and predict the input behavior of the user in the input process of a single word / vocabulary, improve the accuracy of the input result and reduce the number of input operations and correction operations. Increase input speed. An example of a specific calculation method is as follows:

When the user types the N + 1th character of a word / vocabulary, the N + 1th input result is combined to predict the N + 1th user input intention. First calculate the probability that each key on the keyboard may be the user's input intention. After the user enters key A at the N + 1th time, the probability of a single key is calculated as follows:

Among them, P (KI) is the probability that the user intends to press the button I.

Is the drop point model factor, that is, the probability that the user enters key A when he wants to enter key I.

Is the language model factor, that is, in the dictionary, all words with a length greater than or equal to N + 1 and the first N characters are the same characters that the user has entered, and the N + 1th character is the probability of pressing the key I.

After calculating the probability values of all keys, the keys are sorted. The key with the highest probability is the key that the user expects to type, which is the optimal character described below.

In step S120, the optimal character is determined according to the input probability value of each character, and the voice of the optimal character is played.

In one example, in determining the optimal character, in addition to considering the input probability value of each character, the touch-down position is also considered. Specifically, for example, determining the optimal character includes: determining a character with the highest probability value as a candidate optimal character according to the input probability value of each character; and attempting to update the keyboard layout according to the touch-down position and the determined candidate optimal character to Make the candidate optimal character at the touch-press position and the updated keyboard layout meet predetermined criteria; if the updated keyboard layout is successful, determine the candidate optimal character as the optimal character; otherwise, use the default standard keyboard layout to determine the most optimal character. The top character is the key at the touched position in this layout. .

In one example, after playing the voice of the keystroke character, the input of the confirmed character is completed in response to detecting a confirmation signal originating from the user.

In one example, the confirmation signal is a signal triggered by a user gesture, such as a signal triggered by a user lifting a finger or a signal triggered by a double-click operation after lifting a finger. However, this is only an example, and the confirmation signal is not limited to a signal triggered by a gesture. In the case of a gesture triggered, the gesture is not limited to a gesture such as raising a finger or clicking after the finger is raised.

The text input method according to the above embodiments of the present invention improves the accuracy of Chinese and English input without visual feedback, and uses voice to interact with the user, which is very suitable for blind people to use. The text input method can be applied, for example, to a computing device such as a smart phone or a tablet computer.

As shown in FIG. 4, in step S210, the coordinate position of the user's finger on the screen is determined in response to a touch input signal from the touch sensor.

In step S220, it is determined whether the distance between the current touch position and the previous touch lift position is less than a certain threshold.

After receiving the user's contact event, by judging whether the current touch position is close to the previous touch position, the close position indicates that the user will next enter the same letter or the character next to the last key input. The keyboard layout should be updated. For example, if the two input positions are basically the same, the current touch is considered to have occurred on the keyboard layout when it was last lifted, and the keys to be input are the same as the last time. For example, in "hello", the last time you entered "L", and the current touch position is still "L", the two inputs are considered to be related, and there is no need to update the keyboard layout. After the update, it may lead to the inability to enter L correctly; another This situation is basically a key difference between the two input positions. At this time, it is considered that the blind user wants to input the adjacent letter of the last entered letter. For example, the blind user wants to enter "bed" and the last time "e" is input. The position is the position below the last touched position, it is considered that the blind person wants to enter "d" this time. This is also the reason that the keyboard layout should not be updated. After the same update, it may be As a result, it cannot be entered correctly.

Therefore, if the determination result of step S220 is yes, that is, the movement range does not exceed a certain threshold, then proceed to step S240, and restore the keyboard layout to the default standard keyboard layout; otherwise, the determination result of step S220 is no, that is, the user's finger moves relative to the last clicked position. If it exceeds a certain threshold, the process proceeds to step S230.

In step S230, an input character of each character is calculated based on the touched position. For specific implementation of step S230, reference may be made to the description made in conjunction with step S120 in FIG.

In step S250, the candidate optimal character is determined according to the input probability value of each character.

In step S260, an attempt is made to update the keyboard layout as appropriate so that the candidate optimal character is at the touch position, and it is determined whether the keyboard layout meets a predetermined standard.

When the determined candidate optimal character is not the character (key) corresponding to the position touched by the finger on the current keyboard layout, an attempt is made to update the keyboard layout so that the character corresponding to the finger position is the candidate optimal character.

Updated keyboard layouts here include, for example, mobile and / or compressed keyboards. In this example, the interaction between the soft keyboard and the user is adjusted by moving and zooming the soft keyboard, keeping the user's memory of the relative position of the keyboard during the key adjustment process, and improving the efficiency of the user's input adjustment.

Moving and / or compressing the keyboard may include: performing branch or overall movement and compression; compressing the keyboard in a linear or exponential manner with the touch position as the center, and the linear or exponential manner is user selectable.

After the system makes a request to update or adjust the soft keyboard, it will determine whether the request is legal, in other words, whether it meets the predetermined criteria. The criterion for this judgment can be: the movement must not exceed the predefined effective range; the compression must not make any key size smaller than the size threshold that the user can touch, such as whether any key size is smaller than a certain percentage of the original size. If such a key exists later, the request is illegal and no adjustment will be made. If the request is valid, the keyboard will actually be adjusted. The adjustment method is that the entire keyboard is scaled and translated linearly, so that the center of the target key is adjusted to the target position. Here, the movement must not exceed a predefined effective range, which can be determined by using, for example, an input range of a current commonly used input method as a reference.

As shown in FIG. 4, if it is determined in step S260 that the keyboard layout meets the predetermined standard, proceed to step S280, that is, use the updated keyboard layout, confirm that the candidate optimal character is the optimal character, and play the optimal character, otherwise proceed to step In S270, a default keyboard layout is adopted, and the key corresponding to the touch position under the default keyboard layout is determined to be the optimal character, and the optimal character voice is played.

Figures 5a and 5b show examples of soft keyboard movement and zooming processes in actual use. This is the scene of entering the fourth pinyin letter when typing the word "PENGYOU". The actual touch key is "F", the algorithm predicts that the user intends to type "G", and the algorithm requests the keyboard to move the center of the G key to the current touch point. The keyboard confirms that this movement is legal, and finally realizes the movement of the soft keyboard. The resulting voice feedback is "G". The dark gray circle on the letter G in FIG. 5b indicates the finger touch position. Similarly, the dark gray circles in Figs. 7b, 7c also indicate the finger touch position.

FIG. 6 shows a schematic example of illegal movement of a soft keyboard in actual use. Specifically, when the last letter s of the company is input, since the key position is too left, compressing the key will cause the a key width to be smaller than the minimum value (for example, 50% of the original key width), so the movement is illegal. Although the probability maximum key bit is s, because the movement is illegal, the final input key is the key a corresponding to the pressed position.

Returning to FIG. 4, in step S290, the event is continuously monitored. If a mobile event is received, voice feedback is given during the movement. For example, after the system gives a voice prompt for the optimal character, the user thinks that this is not the character he wants, at this time the finger will continue to move to find the remaining characters, and in the process, the voice feedback will continue to be given.

In step S291, in response to receiving the user lifting event, a gesture judgment is performed, and a voice prompt is given.

In one example, operations such as deleting, confirming word selection, switching candidate words, and clearing are implemented through gesture operations.

Determining whether a gesture input has occurred includes detecting contact trajectories in a full screen range instead of being limited to the area occupied by the soft keyboard; using the displacement-time relationship of the contact trajectories to determine whether gesture input has occurred and determining the type of gesture, so that the Type of gesture while determining the corresponding input operation.

In one example, it is determined whether the user has performed a gesture operation through the following steps: When the user completes the lifting motion, if the duration of the entire motion is less than, for example, 200ms, and the displacement in the vertical or horizontal direction is greater than, for example, 70 pixels, and greater than another The displacement in one direction determines that the user has performed a gesture operation. It should be noted that the gesture operation here does not include the operation of the finger moving in the touch state, that is, it does not include the action of searching for the finger on the screen. One-finger or two-finger operation can be determined by the touch point.

In one example, the gesture type may include: single-finger swipe left, two-finger swipe left, single-finger swipe up, single-finger swipe down, single-finger swipe right, and so on.

Exemplarily, the correspondence between gestures and operations can be as follows:

Single-finger left stroke: In the Chinese multi-step word selection, the previous word selection is canceled. In other cases, the last pinyin or letter of the currently selected word is not confirmed. If not, the last Chinese character or letter that completed the word selection is deleted.

Single-finger right swipe: Confirm the input of the current candidate word, and the screen will be automatically displayed after the Chinese multi-step word selection is completed.

One-finger swipe up: Set the currently highlighted candidate as the previous candidate in the list.

One-finger swipe down: Set the currently highlighted candidate as the next candidate in the list.

Two-finger swipe left: Clear all current input.

In one example, after the gesture type is determined, corresponding voice feedback is given, for example, when it is determined that the two-finger left-swipe gesture is performed, the voice prompt "clear all current inputs".

In step S292, the operation corresponding to the determined gesture action is performed, and the information related to the lifting action is also recorded under the clearing determined as the input operation.

For example, when it is determined that the user has performed a lifting action corresponding to the input of a character with audio prompts, the character is input, and the related information of the current lifting is recorded for comparison of the next input.

When a subsequent touch event is received from the user, the process starts from step S210 again.

According to the text input method of the above embodiment of the present invention, the user input is predicted by combining the language model and the drop point model, and the input without visual feedback is adapted by moving and zooming the soft keyboard, using voice to interact with the user, and using gesture operations to complete the text Word selection, confirmation, deletion, and clearing operations during input can improve the accuracy of Chinese and English input without visual feedback and the speed of Chinese and English input without visual feedback.

It should be noted that the keyboard layout is not limited to a visually visible keyboard layout, and may also be an invisible keyboard response area setting capable of achieving the same effect, which has the same effect for blind people.

Taking the input hello as an example, a complete process of inputting a word using the text input method of the embodiment of the present invention is given below.

Figure 7a shows the keyboard state before the user presses a key.

1. After the user presses in the keyboard area, the language model and the drop point model will be used to calculate the key with the highest probability that the user expects to press, and the center of the key will be moved to the user touch point by zooming and panning the keyboard. At the same time, the voice feedback content is the current key, such as 'H'. (The prediction order is displayed on the keyboard through the corner mark, for debugging and display use, non-feedback content)

Figure 7b shows the layout of the keyboard after the user first presses a letter when entering a letter.

2. If the key is not the key that the user desires to input, the key can be retrieved by moving the finger while keeping the finger pressed. During the retrieval process, the keys passed by the user's finger will be feedback in real time through voice, and the ongoing voice feedback will be interrupted by the new voice. During retrieval, the keyboard layout is maintained.

3. After the user confirms the key they want to input through voice feedback, confirm the input by raising their finger. At the same time, the keys input by the user, such as 'H', are voice feedback. After the user types, the keyboard layout remains.

4. Repeat the process from 1-3 to enter the remaining letters of 'Hello'.

5. The user can use gestures to delete the entered letters during word input. With the left stroke operation, you can delete the last letter entered. At the same time, the voice feedback: "X has been deleted" (X is the deleted letter).

Fig. 7c shows a situation where an error occurs when "o" is input. In the example, "i" is input by mistake and deleted with a left swipe gesture. At this time, the voice feedback: "i has been deleted".

Here, the content displayed on the screen in this example will be described with reference to FIG. 7c, but it should be noted that this is only an example, and this is mainly a display of verification experiment results, and what can be done on the screen when it is actually applied to a blind input service Neither display, you can display some content, or you can display different content. In FIG. 7c, the reference numeral 1 drawn by the horizontal line indicates the content currently input according to the embodiment of the present invention, that is, the content confirmed by the user. In this example, the input is confirmed by the user's lifting action, and the reference numeral 2 indicates according to the embodiment of the present invention. The content predicted by the algorithm, reference numeral 3 indicates the input content without running the embodiment of the present invention, that is, the content input by touching the touch point according to the default layout. The number 4 indicates that the currently entered character is i, and the number 5 indicates candidate words, including heli, hellish, helling, and hellip. In addition, the numbers displayed on the characters of the soft keyboard shown below indicate the input probability calculated by the probability ranking. Five characters, that is, the character with the first probability of input is I, the character with the second probability of input is O, the character with the third probability is U, the character with the fourth probability is J, and the character with the fifth probability Is Y.

6. When the user finishes inputting a word, use the right swipe gesture to confirm the input word. At the same time, the voice feedback "Hello has been entered".

In addition, if no letter is entered, the right swipe operation will type a space, and at the same time, the voice feedback "space".

In addition, the user can use the up and down gestures to select the previous and next candidate words, and at the same time, the voice feedback changes the candidate words. This function is less useful in English input, mainly used when there are homophones in Chinese input.

In addition, the user can use a two-finger left swipe gesture to clear all input. At the same time, the voice feedback "empties".

In addition, the user can use the left swipe gesture of the finger to delete the character just entered. For example, FIG. 7d schematically illustrates the deletion of “o” using the left swipe gesture. At this time, the voice feedback: “o has been deleted”, the circle in FIG. 7d And the left arrow gestures to the left.

The input method of this embodiment is particularly suitable for use by blind people, but of course it can also be used by people with normal vision.

It should be noted that the previous gestures and corresponding operations are merely examples, and other gestures may be designed and corresponding to different operations as required.

The input method of the embodiment of the present invention can be applied to various computing devices, such as a smart phone, a tablet computer, and the like.

Embodiments of the present invention may also be implemented as a computing device, including: a touch screen with a touch sensor capable of detecting a touch input; a controller and a memory, the computer-executable instructions stored on the memory, and when the touch soft keyboard is activated: The computer-executable instructions, when executed by a controller, are operable to perform the input method described above.

Embodiments of the present invention may also be implemented as a computer-readable storage medium having computer-executable instructions stored thereon. When the touch-type soft keyboard of a computing device has been activated: when the computer-executable instructions are executed by a computer, Operable to perform the input method described above.

The embodiments of the present invention have been described above, the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the various embodiments described. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

An input method using a touch-type soft keyboard in a computing device includes:

Determining the coordinate position of the user's finger on the screen in response to a touch input signal from the touch sensor;

Calculate the input probability of each character according to the touched position;

Determine the optimal character based on the input probability value of each character; and

Play the voice of the best character.
The input method according to claim 1, the determining an optimal character comprises:

Determine the character with the highest probability value as the candidate optimal character according to the input probability value of each character;

Try to update the keyboard layout according to the touch-down position and the determined candidate optimal character, so that the candidate optimal character is at the touch-down position and the updated keyboard layout meets a predetermined standard;

If the keyboard layout is updated successfully, determine the candidate optimal character as the optimal character;

Otherwise, the default standard keyboard layout is used, and the optimal character is determined as the key at the touched position in this layout.
The input method according to claim 2, further comprising:

In response to the user continuing to access other key characters without lifting the finger, the voice of the key character corresponding to the finger position is played.
The input method according to claim 1, further comprising:

After playing the voice of the key character, the input of the confirmed character is completed in response to detecting a confirmation signal from the user.
The input method according to claim 4, the confirmation signal comprises:

The signal triggered by the user lifting the finger or the signal triggered by a double-click operation after lifting the finger.
The input method according to claim 1, wherein determining the input probability of each character comprises:

Calculate the probability of each character currently entered by the user based on the character string that has been entered and the position that the user touched and pressed, using a combination of the landing point model and the language model.
The input method according to claim 1, before the calculating the input probability of each character according to the touch position, further comprising:

Determine whether the distance between the current touch position and the previous confirmed touch input position is less than a certain threshold,

If so, the keyboard layout at the time of the previous confirmation input is directly used, and the keyboard layout is not updated;

If not, restore the keyboard layout to the default standard keyboard layout, then do probability calculations and determine the optimal characters, and update the keyboard layout as needed based on the results.
The input method according to claim 2-7, wherein the keyboard layout is an invisible keyboard response area setting capable of achieving the same effect.
The input method according to claim 7, wherein the landing point model is a normal distribution.
The input method according to claim 2, wherein said updating the keyboard layout comprises moving and / or compressing the keyboard.
The input method according to claim 11, the mobile and / or compressed keyboard comprises:

Perform branch or overall movement and compression;

Compress the keyboard linearly or exponentially with the touched position as the center;

Movement must not exceed a predefined valid range;

Compression must not make any key size smaller than the size threshold.
The input method according to claim 1, further comprising determining whether a gesture input has occurred,

The determining whether a gesture input has occurred includes:

Detect contact trajectories in the full screen range, not limited to the area occupied by the soft keyboard;

Use the displacement-time relationship of the contact trajectory to determine whether gesture input has occurred and determine the type of gesture;

A corresponding input operation is determined based on the determined gesture type.
The input method according to claim 13, the gesture types and corresponding input operations include:

Single-finger swipe left: delete the last character typed;

Two-finger swipe left: Clear all typed characters;

Single-finger swipe up: Switch the highlighted candidate to the previous candidate;

Single finger swipe down: Switch the highlighted candidate to the next candidate;

One-finger right swipe: Confirm input of currently highlighted candidate
The input method according to any one of claims 1 to 14, further comprising:

Provide audio feedback for touch operations, including:

Feedback on key presses;

Feedback on confirmation inputs;

In response to gestures.
A computing device includes:

Touch screen with touch sensor capable of detecting touch input;

A controller and a memory, where computer-executable instructions are stored on the memory, and when the touch soft keyboard is activated: when the computer-executable instructions are executed by the controller, they are operable to execute the claims 1 to 15 Input method.
A computer-readable storage medium having computer-executable instructions stored thereon, in a case where a touch soft keyboard of a computing device has been activated: when the computer-executable instructions are executed by a computer, they are operable to execute claims Input method described in 1 to 15.