If you feel that promises about speech recognition technology have yet to be fulfilled, you are not alone. Although the usefulness of today’s speech recognition technology in integrated voice response systems (IVR) is undeniable, many customers remain frustrated with the never ending series of prompts and command sequences they must endure in order to navigate through these systems.

Using concepts in speech recognition is a major improvement over current state of the art technology. Conceptual speech recognition goes beyond sound matching and, instead, derives meanings as recognition proceeds. Just imagine an automated call, as in the airline industry, that prompts the user with "Hello! What can I do for you?". The user could as well respond by "I’d like to speak with a representative about paying the flight I reserved" or "I would like to know if flight six hundred arrived and at which gate". Either way, the system would provide the exact response expected by routing the call or providing the flight schedule information that was requested without the intervention of an operator. Following more than three years of research and development on the topic, Conceptual Speech Technologies, LLC has invented and patented a process that allows such an interaction to be possible between a speaker and an automated device.

The bases of this groundbreaking technology are the following. Speech recognition, as performed by humans, has many essential elements related to it. The medium of which being sounds produced by a speaker and heard by a listener. By nature, sound is the carrier of speech, sound is not speech. The listener transforms those sounds perceived into potential words heard. But, because of the impure nature of sounds that is affected by noise, senses and the plurality of interpretations when recognition is limited to pronunciations, humans can not immediately infer the exact words spoken without the help of some other key elements in speech.

Word sequences are not produced randomly by the speaker. They have to respect syntactic rules. The listener follows the same syntactic rules as the speaker to structure a concept inferred from the potential words heard sequenced into a valid syntactic organization. Consequently, syntax needs to be an essential element related to speech.

The goal of speech is to communicate concepts, and syntax is an accessory to that purpose. As an example of that fact, think about the sequence of words “Red rounded house eats a squared ship”. This is not valid speech since, although the sequence of words follows all syntactic rules related to the English language, it fails to communicate a concept. As such, humans perform some type of conceptual analysis to extract a concept from speech whenever it is possible.

In summary, there are three essential elements in speech, and each of them has a specific role:

1. Sound - the medium
2. Syntax - the process
3. Concept - the goal

Conceptual speech recognition is a revolutionary approach that analyzes speech in such a way that all key elements are taken into consideration while being assigned their just role (sound is the medium, syntax is the process and concept is the goal of conceptual speech recognition).

Conceptual speech recognition, in a command and control environment such as in an IVR system, achieves the following objectives:

• Speakers can use their own words to communicate an inquiry.
   
  
• Speakers can ask for multiple things in a single inquiry and get a correct response for each part.
   
  
• Accuracy is significantly improved.
   
  
• Calling durations are reduced.
   
  
• A process instead of a human being can handle calls to a higher level of complexity.
   
  
• The caller opt-out rate is reduced.
   
  
• The system can be used to produce responses that respect the mood of inquiries (polite inquiries, 'yes or no' inquiries or other).

Examples The most immediate expression of conceptual speech recognition technology is expressed through, but not limited to, command and control systems. As an example of the possibilities achievable through this revolutionary technology, the following two inquiries processed by an Airline Response System using conceptual speech recognition demonstrates a typical result: "Welcome to the airport scheduling response system, what can I do for you?" Q: When and where did flight six hundred arrive from New York and how late was the departure of flight three twenty? A: United Airlines flight 600 arrived from New York at 2 32 PM at gate B 2 of Dallas Fort Worth International Airport in Dallas. American Airline flight 320 left at 2 35 PM and was late by 1 hour 15 minutes. The response to this first complex inquiry was calculated in less than one second by Conceptual Speech Commander. The inquiry was spoken using words of speaker's choosing and was not limited by any grammar. Furthermore, any response respects the mood of its inquiry. As in the following example: "Welcome to the airport scheduling response system, what can I do for you?" Q: Could you please tell me how many minutes before flight six hundred will arrive in Dallas? A: Certainly. United Airlines flight 600 already arrived in Dallas 20 minutes ago. The previous inquiry was polite. Therefore, in order to provide a better interaction, the response was produced so that it matches the mood of the inquiry by being equally polite (starting with "certainly"). Furthermore, Conceptual Speech Commander detected that inquiry assumed flight 600 was expected to arrive in a future time. However, since flight 600 already landed, the answer produced by the system corrected the user for his or her wrong assumption through the added word "already" ("United Airlines flight 600 ALREADY arrived in Dallas 20 minutes ago"). This response was calculated within 600 ms. The following example is easily achievable in a customer service IVR application using Conceptual Speech Commander: "Welcome to Connect Phone Company, what can I do for you?" Q: I'd like to speak to a representative about paying my bill with a credit card. A: Thank you. Please hold. You are being transferred to a billing representative. In one easy step, the customer's call has been routed to the correct point in the system. This customer has been placed in queue for the next available billing representative, without having to endure several minutes of navigating through voice prompts. This saves time while providing a level of customer service not available in current IVR systems. The result is a satisfied customer.

Based on the principle that none of us is as strong as all of us, this and other information are available for you to read, and critique, complement or debate from this website.