Introduction to the OpenEars™ Platform
This is a quick overview on the OpenEars™ Platform: the individual parts it is made up of, and how they come together to help you make awesome speech-recognition apps.
The OpenEars™ Framework
OpenEars™ is a free SDK for offline speech recognition and text-to-speech with great free support and lots of documentation, that lets you add speech recognition and TTS to your app in just five minutes using the Politepix Tutorial Tool. It is also a plugin platform that lets you drag-and-drop new speech features into your app using the Politepix OpenEars™ Plugins, and supports speech recognition in seven languages.
The OpenEars™ Plugins
RapidEars is a plugin for OpenEars™ which adds the new ability to do real time speech recognition on in-progress speech for live voice recognition on the iPhone — no waiting for a pause in speech before it can be processed, and no network connection needed. RapidEars is responsive enough that it can be used as a gaming input, and is great for any application that needs real time speech recognition and fast feedback and has a moderately-sized vocabulary.
NeatSpeech is a plugin for OpenEars™ which gives it the ability to use fast, higher-quality voices which have little to no lag even when speaking extremely long statements. NeatSpeech handles all multithreading and queueing in the background so you can just keep feeding it text and it will just keep talking.
RuleORama is a plugin for OpenEars™ which creates rules-based grammars using the same API as OpenEars’ dynamic grammar generation, but the grammars are fast enough for RapidEars to use and are faster even when used with OpenEars™. RuleORama lets you add rules-based recognition grammars to your realtime recognition applications.
Rejecto is an out-of-vocabulary rejection plugin for OpenEars™. A general issue in offline speech recognition is words and noises being heard which aren’t in the app’s vocabulary and the app trying to interpret them as words that it knows. Rejecto improves accuracy by recognizing and rejecting sounds that it doesn’t know.
SaveThatWave is a plugin for OpenEars™ which adds the new ability to record WAV audio files from speech perceived by PocketsphinxController without blocking and to receive notifications of successful write-out of the audio file including the path of the recording. It also wraps up deleting one or all of the recorded files and includes the logic for handling suspending/resuming and recording. Great if you also want to upload speech to a cloud service.