top of page

Emotions: Red- Angry , Blue- Normal , Yellow- Silence

Emotion Detection from Speech

Emotion is judged based on Facial expressions, text for spoken words and from speech tone. The salient features are listed below.
  1. Our emotion detection system extracts emotion directly from speech. The system developed is language independent and accent independent.

  2. As for each call the agent and customer pair will change, the training for the system is required to be done on the fly.

  3. It detects angry, aggressive, happy, sad, fear and surprise mode. The provision is also made for detecting stress mode for forensic applications

  4. It automatically detects gender and change of gender if on client side or agent side a speaker hand over occurs.

  5. It has ability to handle multiple speaker overlapping segments.

  6. For stereo audio files, the system first separates the channels and applies emotion detection on each channel separately.

  7. For Mono files, the speaker’s talks are required to be separated first using speaker diarization system.

bottom of page