Speech Recognition

[laurentlb]

I've built a prototype using the speech recognition browser API. It works, but there are some things to improve:

• Detect when a browser doesn't support the API
• Manual mode: click on an icon each time you want to speak
• Stop the speech recognition when the user clicks on something else (e.g. next, restart the story...)
• Animate the microphone icon when it's actively listening
• Disable speech recognition when audio is playing (otherwise the microphone may capture the TTS)
• Make possible to split a sentence into multiple chunks (combine the outputs, and see if the user said the full sentence). This would help with handling longer sentences.
• Collect feedback. If anyone sees this message, please let me know how the feature works for you.

Note that the browser API is experimental. Not all languages/browsers are supported.