Gaze based text input systems and methods

1. A method for entering text into a computing device using gaze input from a user, wherein the method comprises:

causing a display device to display a text entry interface and a visual representation of a plurality of letters;

receiving gaze information identifying a movement of the user's gaze on the visual representation;

recording an observation sequence of at least a first observation event and a first type of the first observation event based on the movement of the user's gaze on the visual representation, wherein the first type is associated with assigning a character a locked status and assigning the locked character a probability of 100%, wherein the probability is the likelihood that locked character matches the user's intended character; and

providing the observation sequence to a decoder module, wherein the decoder module determines at least one word representing an estimate of an intended text of the user based on the observation sequence and the locked character.

2. The method of claim 1 , further comprises recording a second observation event and a second type of the second observation, wherein the first type is different than the second type, and wherein the record is provided to the decoder module, and wherein the decoder module determines the at least one word from the record based on the first type and on the second type.

3. The method of claim 2 , wherein the first type indicates that the first observation event corresponds to a first fixation for less than a predetermined amount of time, and wherein the second type indicates that the second observation event corresponds to a second fixation for more than the predetermined amount of time.

4. The method of claim 2 , wherein the first type indicates that the first observation event corresponds to a first gaze input, wherein the second type indicates that the second observation event corresponds to a second gaze input and to a non-gaze input.

5. The method of claim 1 , further comprises recording a second observation event and a second type of the second observation event in a record that comprises the first observation event and the first type, wherein the second observation event corresponds to a second letter, wherein the second type is associated with a second letter probability, wherein the record is provided to the decoder module, and wherein the decoder module determines the at least one word from the record based on the first letter probability and the second letter probability.

6. The method of claim 5 , wherein the first letter probability is higher than the second letter probability based on the first type and the second type, wherein the at least one word comprises the first letter and an updated second letter based on the first letter probability and the second letter probability.

7. The method of claim 1 , wherein the first type is indicated by the user fixating upon a character for any amount of time and speaking a triggering word during the fixation.

8. A system comprising:

one or more processors; and

one or more non-transitory computer readable storage media comprising instructions that, upon execution by the one or more processors, cause the system to:

display a text entry interface and a visual representation of a plurality of letters;

receive gaze information identifying a movement of a user's gaze on the visual representation;

record an observation sequence of at least a first observation event and a first type of the first observation event based on the movement of the user's gaze on the visual representation, wherein the first type is associated with assigning a character a locked status and assigned the locked character a probability of 100%, wherein the probability is the likelihood that locked character matches the user's intended character; and

determine at least one word representing an estimate of an intended text of the user based on the first observation sequence and the locked character.

9. The system of claim 8 , wherein the at least one word is determined based on a language model, and wherein the execution of the instructions further cause the system to:

receive a correction to the at least one word based on a correction interface that displays the intended text and an insertion icon between the two words of the intended text, the insertion icon being a place to gaze at between the two words to initiate an insertion of a word at the place between the two words, wherein the system receives user gaze information indicating a selection of the insertion icon; and

update the language model based on the correction.

10. The system of claim 9 , wherein the correction comprises a corrected word, and wherein the execution of the instructions further cause the system to:

determine that the at least one word comprises at least a locked letter of the one or more locked letters; and

store the corrected word in the language model based on the at least one word comprising at least the locked letter.

11. The system of claim 10 , wherein the execution of the instructions further cause the system to output a predicted string that comprises the at least one word, and wherein the execution of the instructions further cause the system to generate a confidence measurement based on a number of locked letters and on corrections made to the predicted string.

12. The system of claim 11 , wherein the corrected word is stored in the language model based on a comparison of the confidence measurement to a confidence threshold.

13. The system of claim 11 , further comprising: updating a frequency of adapting the language model to use corrected words based on the confidence measurement.

14. One or more non-transitory computer readable storage media comprising instructions that, upon execution on a system, configure the system to perform operations comprising:

displaying a text entry interface and a visual representation of a plurality of letters;

receiving gaze information identifying a movement of a user's gaze on the visual representation;

recording an observation sequence of at least a first observation event and a first type of the first observation event based on the movement of the user's gaze on the visual representation, wherein the first type is associated with assigning a character a locked status and assigned the locked character a probability of 100%, wherein the probability is the likelihood that locked character matches the user's intended character; and

determining at least one word representing an estimate of an intended text of the user based on the observation sequence and the locked character.

15. The one or more non-transitory computer readable storage media of claim 14 , wherein the operations further comprise recording a second observation event and a second type of the second observation event in a record that comprises the first observation event and the first type, wherein the at least one word is determined from the record based on the first type and on the second type.