chen-ye / bespoken Goto Github PK

Currently, there is no way to save / load presentations. Having a database backend that saved presentations would be incredibly useful.

I think a backend based in Node.js and MongoDB would be the best, as that would complete the standard "MEAN Stack" and the relatively unstructured data of a Bespoken presentation seems to fit the MongoDB model the best (compared to a relational database, where the advantage is in normalization with foreign keys and etc).

I might get around to doing this, but it would be great to have some help.

In terms of using the app, there are only the instructions in the blue box.

While the instructions are good enough to show off the features of the app, a paragraph description of the app's use case might work better. The best case would be an interactive tutorial that coaches the user through speaking into their computer.

Modern, well presented presentations rely on graphics. Images, charts, and typography assist in illustrating an author's point. We should have the ability to generate charts on the fly - relying on a framework such as D3 may help.

The Google Web Speech Recognition API demo has a pretty good error handling system, which I--in the haste of PennApps--neglected to include when coding up our version.

There are a series of errors that can occur with the Google Web Speech Recognition API (such as a user trying to use Bespoken from an unsupported browser), as well as potential errors in other places of the JavaScript code.

Error-handling functionality is a must-have for Bespoken.

I've talked about the keyboard controls of Bespoken, and some people are concerned that the keyboard controls could be distracting to the presentation.

While the original PennApps demonstration involved a LEAP Motion controller, the controller was pretty problematic and would not be a realistic option for presentations anyway.

It would be really interesting to have a presenter practice a presentation, using the keyboard to switch slides, and then the presentation software learning to switch the slides automatically when the "transition words" between the end of one slide and the beginning of another are spoken.

If this was accurate, it would be an extremely smooth method of presenting. However, doing this accurately could prove to be nontrivial...

The second least sexy alternative could be to advance the slide based on a static, easy to recognize keyword like "Next!"

The absolute least sexy alternative is keyboard or mouse navigation. We could make this better by customizing the keyboard or mouse signals captured, allowing people to customize Bespoken to the keys sent by their PowerPoint clickers, but it would still be unwieldy in most use cases.