Voice4id is the product of Voisis for professional applications of Speaker Recognition. The system provides for two different phases of operation:
- in the training phase, the system is prepared to extract the “footprints” of all the known entries. The duration of the audio required for training depends on the accuracy to be obtained from the system. A few minutes for each person are sufficient for a high level of accuracy. This step can be repeated as new items are added or deleted.
- During the recognition phase, the audio signal, in the form of files or streaming, is sent to the system, and the voice is compared with the set of items “notes.” Collected enough information, the system responds with what is the identified speaker (or speakers to it more like) with possible scores of verisimilitude. The longer the duration of this phase and the lower the percentage of error. A few seconds are enough for a high level of accuracy.
The system is easily integrated into any application through an interface ‘Web Service’ (REST, SOAP). MAIN PRODUCT FEATURES:
- Training of known entries schedulable.
- Analysis of files o streaming audio.
- Text-independent: it is not necessary for the user rule slogans or phrases.
- Scalability of the system according to the number of users.
- Interface’Web service‘ for easy integration into any application.
- Ability to use different authentication procedures.
- Administration and monitoring of the system through the browser.
- The system is normally supplied as a stand alone device or virtual server .
Voice4id is also usable in combination with other products Voisis:
- VoisisControl: the recognition of voice commands and gestures is coupled with voice recognition
- VAMP: in an archive (audio-video) can be analyzed for known entries.