> Phonexia Speaker Identification System Demo

Phonexia Speaker Identification System Demo

The application searches for a speaker in an archive of speech records. The records are sort according to similarity with the target speaker. The application integrates also gender identification.

sid-app screenshot1 sid-app screenshot2

Download our demo application now!

Download command line version of the demo application now!

>  Technology:
  • Uses the voiceprints from the sounds. Each person has unique shape of vocal cavities and the way of moving mouth.
  • Is language, text and channel independent.
  • Applies the state-of-the-art channel compensation techniques, verified in NIST evaluations.
>  Key Features:
Input formats: WAV or RAW with linear coding (16 or 8 bits), A-law, Mu-law
Training input: Speaker model is created from one or more recordings
Test input: Directory with audio files
Output: List of recordings sorted according to similarity to the searched speaker. The list can be post-processed, filtered and exported
Sampling frequency: 8000Hz
>  Use:
  • Software is an example demonstrating a speaker search
  • Get more by accessing directly Phonexia libraries - allows for easy integration with your intelligence-oriented application
>  Consultation:
  • Tuning system parameters for the best performance on customer's data
  • Adapting or designing the system to your specific task