This is a potted story of, what was for me, a pretty big undertaking. It was my biggest ever electronics project, and kept me off the streets (on and off) for 18 months.
Before I undertook this project, I had no experience of laying out circuit boards professionally. Until then, everything I did was either built on vero-board, or at best, etched copper-clad board, immersed in ferric chloride. In many ways this hobbyists way of producing circuit boards is pretty heavy-going, and the results are never stunning.
I was in desperate need of a serious project, as I had gained a lot of experience with Microchip's PIC devices, with nothing to do with it. In casual conversation with the workshop manager of the company I work for, it came out that one of our customers was thinking about having a speech system on an organ. This organ is at an international school for organists, and they wanted to enhance the experience for visually challenged organ students.
I took this as my cue, and did some research into the availability of digital speech recording devices. There were a fair abundance of limited speech-time devices, but I was looking at quite a few minutes in total. I ended up with some devices which can store a maximum of 16 minutes. The talking system has two of these devices, so there is the potential for 32 minutes if needed. At the sampling frequency I eventually settled on, the production system has somewhat less than that now - still a serious amount though! The devices used can, the makers claim, store the speech patterns for up to 100 years. They can also be re-recorded on 100,000 times.
The system, if you set it up to, can give a running commentary on registration changes while using the instrument. The user can alter the way it works through extensive set-up options. At its basic level, it will just announce single stop changes. A useful function, is that the system can tell you what digits are being shown on the organ's display panel(s).