As promised last week, here you can see a video presenting our final prototype:
During the last week’s presentation of our implemented prototype, where all the students from our course tried out our application, we received a lot of feedback and improvement ideas which we implemented for the final presentation on 15th July.
The improvement ideas included a lot of ways to increase the usability of our application. For example by implementing a feature where the cooking guide can be commanded by saying “repeat” out loud to repeat the speech output of the current cooking step or to display all available voice commands for the user by saying “repeat”.
We also improved the performance of our implemented Wolfram API Interface by implementing a caching feature to reduce waiting time during each request to the Wolfram Servers.
The complete change log for our final prototype is listed below:
- say “repeat”: to repeat description of current step
- say “help”: to display all available voice commands
- say “previous” (alternative to back): to go one step back
- Implemented nutrition facts dialog
- Show current cooking progress by showing “Step x / y”.
- Added link (dialog) to index page to get list with all voice commands.
- Added loading sign to indicate processing and retrieval of wolfram data (for nutrition facts)
- Calculate nutrition facts of whole recipe by Wolfram Api
- Added Wolfram Server request caching to reduce waiting time / delay
- several Wolfram API interface bug fixes
Stay tuned for the release of a short clip introducing our final prototype next week!
Last week our group implemented the feature, which was a highlight of our project, namely the Nutritional Information retrieval using Wolfram API.
The implementation time took us longer then we expected, therefore unfortunately we had to postpone the last feature on our agenda – the speech output as a complementary output modality during the cooking guiding steps.
Last week, we prepared our project for the first release of the prototype of our cooking aid application which our group is going to present in our course meeting on Wednesday next week.
We fixed several code issues that our prototype had that were mentioned in our last blog entry. Speech recognition using Annyang and hands free navigation through the application now works flawlessly. Also, the Graphical User Interface was polished.
For the second release of our prototype we started to implement the retrieval of the nutritional information using the Wolfram API and also speech output using Google’s Speech Synthesis API. With the newly introduced speech output, the cooking guide will read the cooking instructions so that the user of the application doesn’t accidentally cuts his hands if he tries to read the shown instructions while cutting a vegetable.
Last week, after the intermediate presentation which was a big milestone in our project, we made some progress with the content of our application, and the application logic as well.
For our first recipe, we decided that it would the best option (and also most legal) to create an original content by cooking something on our own and document the steps using notes and a camera. This resulted in a nice looking meal which our future users will be happy to try out by using the first release of our prototype of the application:
Our application is now able to filter the recipe list by typing a search term inside the textbox above the list. For example by typing an ingredient name, the recipe list is filtered on the fly:
In order to resolve issues of the speech recognition framework we are using where our browser did not save the settings for microphone allowance we have to enable our backend to use https request handling. This way, the recognition engine will work flawlessly without constantly interrupting the user by asking whether he allows his microphone to be used.
Last week our group came up with first drafts of how the final graphical user interface (GUI) of our multimodal application should look. Also we kicked off a project on github where our group already pushed first backend and database implementations.
The Recipe List
The first that the user of the cooking aid application will see is a list of Recipes that are stored inside the database. The List can be queried using the search box located on the top left of the screen. Inside the textbox, the use can type a recipe name, an ingredient name or the meal type and the list is then being filtered accordingly.
After a Recipe was selected from the recipe list, the application changes to the Recipe Overview view, inside which the user obtains all the information he needs about the selected recipe, including nutritional informations and the difficulty of the cooling steps. By Clicking/Touching the “Start Cooking” or by simple saying “Start cooking” into the microphone the application changes to the Recipe Steps View of the application.
Here the user control the application solely with speech input. By saying “Forward” or “Back” he can navigate through Cooking Steps.
Also can receive additional information from the application regarding nutritional information of the ingredients. For example by simply saying out loud “Show me nutritional information about two slices of cheese”, the user gets the asked information display on the screen. This feature will be accomplished with help of the Wolfram Alpha API: