The 4 C's | Corey's home page | Corey's Talking Toaster | Project Update 2

Talking Toaster

Project Update 2

Corin Anderson & Chris Setter

CSE 477, Spring 1996

Abstract

Our project is to design and build a toaster for the Computer Scientist. It's main features will include speech recognition and voice response, heat and other sensors for monitoring toast quality, and automated bread platform. The user may select toast quality from voice commands stored in the toaster's memory.

On first glance, the toaster will look like a toaster of the classic 1950's era, complete with polished aluminum sides and traditional toasting lever. Inside, however, is a completely different story. In an external logic box 1 will be a microcontroller that will process the sensor input and generate appropriate responses.

Sensors include a motion sensor to detect a toast-hungry person; visible light LEDs to detect smoke, indicating burned toast; and a temperature sensor to measure the energy transmitted to the bread while toasting. For the user interface, there is speaker-dependent speech recognition and voice response. This allows for several individuals to have their own custom toast preferences set based upon their personal voice prints. The voice response module will provide friendly feedback to the user based upon the user's selection.

To ensure the safety of the toast connoisseur, the power lines in the toaster are properly insulated. Additionally, to ensure the safety of the toaster, the microcontroller will be shielded from the line current switch by external circuitry. This is the case for the heating coil switch (solid state relay) and motion sensor (relay).

Status

The project is not as far along as we had hoped it would be at this point. Some snags came up that were unexpected that caused delays of as much as a week. We believe that we can still finish the project by the June 7th due date, however.

Issues Resolved

About half of the issues raised in the first project update have been resolved in these past two weeks. Here is a list of those issues:

Voice Playback: The voice playback circuit has been constructed and debugged on the protoboard. Microcontroller code was written that successfully records and plays voice clips at any address on the voice chip. The same microcontroller ports that will be used for the final design were used in this small program to reduce final assembly problems.
Speech recognition: A great deal of time and effort was invested into the speech recognition circuit. The data sheets that were supplied with the chip (Hualon's HM2007) did not have accurate information. Much experimenting was necessary to determine the correct sequence for commands and responses. Even more difficult was the experimenting with the microphone circuit. Quelling the noise from the power supply and the evaluation board was a nontrivial task.
Proximity Sensor: The proximity sensor used is more of an off-the-shelf part. It is the motion sensor from a porch light fixture. There were three reasons that this unit was used versus constructing a pyrometer circuit from scratch. The first is that this unit was known to work. The second is that finding parts for a motion sensor (i.e., finding the pyrometer) is not easy. The third reasons is that this sensor has an RF transmitter and receiver. This allows the motion sensor to be located some distance away from the receiver and toaster, allowing for much easier placement.
Modular Design: One of the design issues we faced early on was the interconnection of our external logic box to the toaster and its sensors and actuators. We wanted to have a reliable connection between the two units but maintain a modular design. The solution was to implement a quick release toaster cover and a socket on the external logic box. The socket strip allowed for the sensor and actuator connections to be made quickly and reliably. We also have a self-coupling motor mount in the toaster. This allows the motor to be mounted in the external logic box and have just the worm gear protrude into the motor.

Issues Pending

We wish there were no major issues still pending, and that we could see nothing but clear sailing from here on in, but that's not the case. There are still some components that have not been tested or experimented with. Fortunately, these components are not as inherently complex as the voice response or speech recognition systems, so experimenting with them should be relatively easy.

Heat Sensor: No experimenting has been done with the heat sensor yet.
Bread Sensor: No experimenting has been done with the bread sensor yet.
Smoke Sensor: No experimenting has been done with the smoke sensor yet.
Bread Platform Actuator: We have obtained a suitable motor with a 5" screw gear to be used to raise and lower the toaster's bread tray. However, the motor was purchased as salvage, so there is no documentation for it. We have contacted the manufacturer, but their documentation has not arrived yet.

Construction and Debugging Plan

All the electronics for the toaster will be housed in a case underneath the toaster. The controlling logic will interface to the toaster and its sensors through a modular plug so the toaster and case may be separated. This will ease the debugging of the design as well as increase the robustness of the entire unit.

There are four major stages of this project. These stages were detailed in the first project update. Below is their current status.

Voice Interface System (VIS): The VIS consists of the voice playback module and the speech recognition module. Experiments have been conducted on each module individually. Version 1.0 of the Toaster Operating System (TOS) is currently being debugged, in which these two modules have been combined into the one VIS.
Sensors and Devices System (SDS): The SDS includes all the sensors and actuators in the project. So far, the only sensor tested and implemented is the proximity sensor. The actuator (the bread platform motor assembly) is in the process of being implemented. The motor will be mounted in the external case and the worm gear will protrude into the toaster. The remaining sensors will be installed in the toaster before they are fully tested. If need be, the sensors will be moved later.
Toaster Operating System (TOS): The TOS is the microcontroller code that unifies the several modules of the toaster. The present version of TOS (v1.0) includes code for the speech recognition and voice response systems, as well as a rudimentary flow of logic that will control the entire toaster.
Debugging the design: The debugging process is an ongoing task. In particular, however, it is more evident in the implementation of the TOS microcontroller code. Even though the code used by TOS has been tested individually by the modules of the VIS, the code as a whole needs debugging to make it all work well together.

Appendix A

Initial Design Schematics

High level design schematic.

Speech Recognition Module schematic

Voice Response Module schematic

Appendix B

TOS Flowchart v1.0

1 The logic box is a metal box with the same footprint as the toaster. All circuitry will be housed in this box to protect the circuitry from the extreme temperatures of the toaster's heating coils.

Corin Anderson | corin@the4cs.com
Last modified: September 8, 1996