carpool

Commit Description

• [VA]: Voice Assistant
• [HE]: Hi Emotion
• [MC]: MicCentral

Initial commit (2021/04/05)

• Added maintenance related files.

[VA] feat: aimybox (2021/04/15)

• Applied Aimybox Framework (framework for voice assistant).

[VA] docs: add readme (2021/04/15)

[VA] feat: add kaldi plugin (2021/04/16)

• Added kaldi plugin for voice trigger. But not working.

[VA] feat: start with recognition mode (2021/04/16)

• Made voice assistant to speak first on activity start.

[VA] feat: read notifications (2021/04/19)

• Added NotificationListener service for reading notifications.

[VA] feat: add NotificationProxy (2021/04/21)

• Added utility function related for pusing notifications.

[VA] refactor: NotificationListener (2021/04/21)

• Moved notification-read related codes on MainActivity to NotificationListener.

[VA] feat: new version (without aimybox) (2021/04/29)

• Implemented new VA without Aimybox Framework. Now uses SpeechRecognizer and TextToSpeech.

[VA] chore: remove old version (2021/04/29)

[VA] add README (2021/04/29)

[VA] feat: vibrate before alerting user (2021/05/04)

• Make phone vibrate before alerting user for unreplied notification.

[VA] feat: add application state (2021/05/16)

• Roughly implemented state machine of application.

[HE] feat: add WavRecorder (2021/05/16)

• Initialized project HiEmotion.
• Added WavRecorder class that records and (will) detects human voice.
• Roughly using ZCR (Zero Crossing Rate) and Energy (sum of data[i]**2) for detection.

[HE] feat: add FFT library (2021/05/19)

• Applied JTransform (FFT) library to eliminate noises on audio for better detection.
• Need more discovery about library & FFT.

update README (2021/05/19)

[HE] feat: replace FFT(DSP) library (2021/05/21)

• Remove JTransform, Applied TarsosDSP which has pitch detection.

[HE] feat: add socket server & emotion model (2021/05/21)

• Add python socket server & MLP model where input=audio features, output=emotion (softmax).

[MC] feat: add MicCentral & integrate with HE, VA (2021/06/01)

• Add MicCentral application.
• Reason for adding MC
  • Android allows only one application to access mic at a time.
  • In this constraint, Hi Emotion would not work while VA is active.
  • Thus, MC is added to propagate mic recorded buffer (file) to other applications.

update README (2021/06/01)