July 2009 Edition
We hope that you are having a great summer!
Voice In Action:
In each newsletter, Voxware will invite readers to submit questions about Voice technology and then we will answer the questions, with the help of our customers. This newsletter's questions are on speech recognition.
Question: What is the difference between a Speaker Dependent and Speaker Independent voice recognizer?
Answer: Speech recognition is the process of converting spoken word into machine-readable input. In other words, it is how a voice solution communicates back and forth with the user. There are two types of voice recognizers: Speaker Dependent and Speaker Independent. A Speaker Dependent voice recognizer requires each user to train their voice to the recognizer. It then creates a voice profile for the user, or voice fingerprint. A Speaker Independent system allows anyone to use the system without training. These typically have a much higher mis-recognition rate, because the voice recognizer is designed to accommodate any voice.
Question: Why is the differentiation important?
Answer: Warehouse environments tend to be very volatile, so a voice recognizer must be able to withstand the harsh conditions of a warehouse, whether it is cold, hot, loud, quiet, etc. For example, Coldstar, a Belgian third-party logistics (3PL) running Voxware's Voice solution with Motorola devices, operates a frozen food DC which runs at -13F. As you can imagine, the equipment used in the DC to control the temperature is very noisy and makes it difficult to hear. Luc Put, CFO of VEHA-Groep, Coldstar's parent, said that Voxware's system performance in a noisy freezer environment has been outstanding. Voxware's Speaker Dependent Voice Recognition engine has twenty-five years worth of development including seven patents.