0.1 The idea  (Page 2/3)

After the session Jim Moore told Paul Breedlove, who had just joined Jim’s team, about the brain storming session. Jim asked Paul to look into the spelling learning aid concept and the possibility of using bubble memory as a storage medium for the spelling lists. They discussed the fact that there was no solution identified to solve the “presentation of the word to be spelled” problem. Paul was intrigued by the concept of the spelling learning aid. He went home that night, had dinner, and wrestled with the problem of how to present the words. Later that evening, his 9 year old daughter and his 7 year old son asked for some help on their spelling assignments. As he called out the words and listened to the answers, he corrected them when they made errors. He decided that the spelling learning aid would have to speak the words to the students, just like he and other parents did all the time. After working for a year in the speech research department at TI, designing electronics to assist researchers in their efforts, Paul had a good understanding of the capability of the researchers and felt they could help solve the problem. This would allow the product to speak the words that were to be spelled in addition to the other responses and instructions the product would need to use, by using synthetic speech technology.

Paul spent the next several days detailing his ideas for the spelling learning aid in his engineering notebook (see Appendix 1). He also soon realized that the spelling learning aid concept would not require a sophisticated memory, like bubble memories, but only needed Read Only Memory (ROM). However, it would need a large ROM, far larger than the state of the art could give. His estimate was that the product would need approximately 256K bits. At that time, the state of the art for ROMs was about 4K bits. The good news was that he had resolved in his mind one of the burning issues – how to store the massive amount of speech data at a very low cost. It turns out, the data could be compressed by synthesizing speech. Now, only the “impossible” speech synthesis task was an unknown.

The “spelling bee”

Paul originally called the spelling learning aid the “Spelling Bee” which became the project name throughout development, Figure 3 shows his initial concept and the cleaned up version is seen in Figure 4. In his engineering notebook he detailed how the product would interact with the user. The fact that speech synthesis had yet to be done in a single chip didn’t seem to bother him a great deal. What he needed was funding to get the program moving forward.

Finding funding

At that time TI had a formal method of acquiring funding called OST (Objectives, Strategies and Tactics). The OST had been put into action by one of our former CEOs, Pat Hagerty. Paul took the proposal through the OST process, but it was felt to be too risky. The OST objective was to make sure all of the projects necessary to create TI's future products would be funded. It allowed business managers to put more focus on the net revenue for the present month, quarter or year while their future products were being developed under OST. After reviewing the concept of the spelling bee, they recommended that Paul submit it for special “Wild Hare” funding as the concept had far more risk than allowed under OST funding. The Wild Hare funding source had been set up for high risk programs with a chance of a big reward if successful, the goal was to fund projects with less than ten percent chance of succeeding. Though the project seemed intriguing it was again deemed too risky, even for Wild Hare money. After reviewing the failure to obtain funding with the division manager, it was suggested that Paul seek IDEA Funding from the IDEA Coordinator, Ralph Dozier. The concept behind the IDEA funding was as follows: