Final report - Atomic Rhubarb

Final report - Atomic Rhubarb
The George Washington University
Department of Computer Science
CS339 Final Project
Project Final Report
Z8 Speaks
Apr. 25, 2006
Duoduo Liao
[email protected]
Table of Contents
Project Abstract................................................................................................................... 4
Project Abstract................................................................................................................... 4
Project Status ...................................................................................................................... 4
1 Design Overview ............................................................................................................. 4
1.1 The Purpose and Requirements ................................................................................ 4
1.2 The Hardware Design ............................................................................................... 5
1.2.1 The hardware block diagram............................................................................. 5
1.3 The Software Design ................................................................................................ 7
2 Specifications................................................................................................................... 8
2.1 Hardware Modules.................................................................................................... 8
2.1.1 Z8 Encore!TM Flash Microcontroller Development Kit..................................... 9
2.1.2 VOICEDIRECT Speech Recognition Kit ............................................................... 9
2.1.3 SpeakJet Speech Synthesis Chip ...................................................................... 12
2.1.4 DS1307 Real-Time Clock (RTC) Module ........................................................ 13
2.1.5 MAX6610 Temperature Sensor........................................................................ 14
2.2.6 Parts List.......................................................................................................... 15
2.2 Software Modules ................................................................................................... 15
3 Implementation & Construction .................................................................................... 16
3.1 Speech Recognition ................................................................................................ 17
3.1.1 VOICEDIRECT Board Stand-Alone Test............................................................. 17
3.1.2 VOICEDIRECT Driver Development for Z8....................................................... 19
3.2 Speech Synthesis..................................................................................................... 20
3.2.1 SpeakJet Chip Stand-Alone Test ...................................................................... 20
3.2.2 SpeakJet Driver Development for Z8 .............................................................. 21
3.3 Real-time Clock Reading/Setting ........................................................................... 23
3.3.1 I2C Protocol Control........................................................................................ 23
3.3.2 Data Transfer ................................................................................................... 24
3.3.3 Time/Calendar Reading/Setting....................................................................... 25
3.4 Temperature Driver and Reading ........................................................................... 26
3.5 Phrase Allophone Editing ....................................................................................... 27
3.5.1 Temperature Allophone Editing....................................................................... 28
3.5.2 Time/Calendar Allophone Editing ................................................................... 28
3.6 System Integration and Main Application Flow Chart........................................... 29
4 Conclusions.................................................................................................................... 31
5 Attachments List ............................................................................................................ 32
2
Table of Figures
Figure 1: The Block Diagram of Hardware Design............................................................ 5
Figure 2: The Schematic for the Design ............................................................................. 7
Figure 3: The Block Diagram of the Software Design ....................................................... 8
Figure 4: Main Hardware Components of Z8 Speaks ........................................................ 9
Figure 5: VOICEDIRECT Board (front and back) ............................................................... 10
Figure 6: VOICEDIRECT Pinout ......................................................................................... 10
Figure 7: SpeakJet Chip (Left) and Pinout (Right)........................................................... 12
Figure 8: SpeakJet Pin Details and Electrical Specifications ........................................... 13
Figure 9: RTC Module (Left) and RTC Module Schematic (Right) ................................ 13
Figure 10: DS1307 Address Map and Timekeeper Registers........................................... 14
Figure 11: MAX6610 Temperature Sensor Configuration............................................... 14
Figure 12: The Final Hardware Connection for the Z8 Speaks System ........................... 16
Figure 13: Stand-Alone Test for VoiceDirect Hardware .................................................. 18
Figure 14: Hardware Connection for VoiceDirect Communication with Z8 ................... 19
Figure 15: Hardware Test for SpeakJet Demo Mode ....................................................... 21
Figure 16: The Schematic for SpeakJet Communication with Z8.................................... 21
Figure 17: DS1307 Data Transfer..................................................................................... 24
Figure 18: The Flow Chart of Z8 Speak Main Application.............................................. 30
3
Project Abstract
This final project report describes in detail how to create a Z8 speaks like a talking robot.
The Z8 Speaks is mainly composed of Z8 microcontroller board, VoiceDirect speech
recognition board, SpeakJet speech synthesis chip, real-time clock, and temperature
sensor. The user can interact with the Z8 Speaks through speech communication based on
speech recognition and synthesis. It has the capabilities to display current temperature,
time, calendar, name, music, reset, and other information on the LEDs in the Z8 board
upon spoken request as well as speaking corresponding information. This project
provides twelve different operations, such as temperature, time, calendar, reset, etc.,
which have been already set well in the program. Before the user talks with the robot, his
or her voice commands need to be trained correspondingly for the preset operations and
stored as speech patterns. Then the user can ask the questions for the robot. If the voice
does not match any of trained voice commands, a speech instruction – “words cannot be
recognized” or “Repeat, look for” will be given to allow the user to try again.
Project Status
The overall work for the final project works very well as planned in the project proposal.
Specifically, the final project meets the proposed requirements as follows,
• Read the temperature from ADC temperature sensor
• Read the Real-Time Clock (RTC) using I2C bus
• Control the speech recognition board using GPIO and hardware switches
• Control speech synthesis through serial data line
• Interact with the user over the speech communication
In addition, more voice operations and speech controls are added into the final project
beyond the proposal as follows,
• Write/Reset the RTC using I2C bus
• Software decoding to reach up to 15 word commands for speech recognition
• Allophone phase editing
• Numerical pronunciation (0 – 69)
• Calendar pronunciation (Jan. 1 – Dec. 31, Monday - Sunday)
• Non-standard language speech synthesis (Chinese, AM/PM)
• Robot music playing
Project Final Demonstration Video (13MB):
http://home.gwu.edu/~dliao/cs339/Liao_z8Speaks_320x240.mpg
1 Design Overview
1.1 The Purpose and Requirements
The purpose of this project is to create a Z8 speaks. It has the capabilities to display
current temperature, real-time date, day, time, name, and other information on the LEDs
in the Z8 board upon spoken request as well as speaking corresponding information. The
4
Z8 Speaks mainly consists of Z8 microcontroller board, VOICEDIRECT speech recognition
board, SpeakJet speech synthesis chip, real-time clock, and temperature sensor. The Z8
microcontroller will control temperature reading, RTC reading/writing, speech
recognition, speech synthesis and output.
The proposed requirements that the project must meet are:
• Read the temperature from ADC temperature sensor
• Read the Real-Time Clock (RTC) using I2C bus
• Control the speech recognition board using GPIO and hardware switches
• Control speech synthesis through serial data line
• Control LED display with corresponding information
• Interact with the user over the speech communication
1.2 The Hardware Design
1.2.1 The hardware block diagram
The block diagram of overall hardware design is shown in Figure 1 as follows.
VOICEDIRECT
Speech
recognition board
SpeakJet
Speech
synthesis chip
GPIO
Serial Data Line
GPIO
Z8
Microcontroller
Microphone
Speaker
I2C
ADC
Real-time Clock
Temperature Sensor
Figure 1: The Block Diagram of Hardware Design
1.2.2 The Schematic for the Design
5
The interfacing of the Z8 microcontroller to the VioceDirect board, SpeakJet chip, RTC
module, MAX6610 temperature sensor, microphone, switches, and speakers is shown in
the following schematic for the design in Figure 2.
(a) The Formal Schematic for the Design (Final Version)
6
(b) The Schematic Scratch (Work Progress)
Figure 2: The Schematic for the Design
1.3 The Software Design
The software is designed for four major modules, Speech Recognition Manager, Sensor
Manager, Speech Synthesis Manager, and Driver Manger. The block diagram of the
overall software design is shown in Figure 3 as follows.
7
Speech Recognition
Manager
Speech Synthesis
Manager
Sensor Manager
Driver Manager
Timer, Switch
Speech
Input
Allophone
Editing
Data Reader
Speech
Recoginition
Speech
Synthesis
Data Display
Speech
Output
Data
Preprocessing
for Speaking
Voice
Command
Determination
SpeakJet
Driver
LED Display
Driver
Temperature
Driver
Real-time
Clock Driver
(Temperature,
time, date,
name, etc.)
Speech
Recognition
Board Driver
Figure 3: The Block Diagram of the Software Design
2 Specifications
2.1 Hardware Modules
The project design consists of five main hardware modules: Z8 Encore!TM Flash
Microcontroller Development Kit, VOICEDIRECT speech recognition board, SpeakJet
speech synthesis chip, Sparkfun DS1307 Real-Time Clock Module, and MAX6610
temperature sensor.
The main hardware components for Z8 Speaks are shown in Figure 4.
8
Wires
Switches, Resistors
VOICEDIRECT
SpeakJet
Microphone
LEDs
Speaker
Real-time Clock Module
Breadboard
MAX6610 Temperature Sensor
Z8 Encore! Development Board
Figure 4: Main Hardware Components of Z8 Speaks
2.1.1 Z8 Encore!TM Flash Microcontroller Development Kit
The Z8 Encore!TM Flash Microcontroller Development Kit allows the user to design and
evaluate projects using the eZ8 microcontroller. The kit contains a Z8F6423 module,
which contains the Z8F6423 device running at 18.432 MHz, with 64 Kbytes of Flash
memory and 4 Kbytes of register RAM. This evaluation board provides 12-channel 10-bit
A/D converter, four 16-bit timers, a watch-dog timer, 60 General-Purpose I/Os (GPIO),
and 24 vectored interrupts. The board also contains SPI, I2C, and 2 UART ports with
IrDA encode/decoder, 3-channel DMA controller, four 7x5 LED arrays, three
pushbuttons, and embedded modem socket. Furthermore, it contains Zilog’s proprietary
ZDS II for debugging and programming. The adapter provides 9VDC for the evaluation
board. The board supports 3.0-3.6V operating voltage with 5V-tolerant inputs.
2.1.2 VOICEDIRECT Speech Recognition Kit
VOICEDIRECT Speech Recognition Kit includes speech recognition board (as shown in
Figure 5), one microphone, one speaker, three microswitches, and two 100KOhms
resistors.
9
VOICEDIRECT is a speaker-dependent speech recognition module, allowing training of up
to 15 words. Using sophisticated speech recognition technology, VOICEDIRECT maps
spoken commands to system control functions. Each time one of the words is recognized,
a corresponding output pin on the module is toggled for one second.
Figure 5: VOICEDIRECT Board (front and back)
Figure 6: VOICEDIRECT Pinout
The module pinout layout is shown in Figure 6 and major module pinout used for this
project is listed in the Table 1 as follows.
10
Table 1: Major Module Pinout
Name
PREAMP IN
Module Pin
JP2 - 1
MIC BIAS
JP2 - 2
AGND
JP2 – 3, 5
-+5V
JP2 - 4
PWM1
JP2 - 6
PWM0
JP2 - 7
DAOUT
JP2 - 8
-RECOG
JP2 - 10
-TRAIN
JP2 - 11
OUT1-OUT7
JP2 – 12-18
HIGH/OUT8
JP2 - 19
Description
Microphone Input
Connection
Mic Bias (Elec.
Microphone)
Analog Ground. For
noise reasons,
analog and digital
grounds should
connect together
only at VOICE
DIRECT.
5 Volt(+) Power
Supply Connection
Pulse Width
Modulator Output 1
(multiplexed)
Pulse Width
Modulator Output 0
Analog Output
(unbuffered)
Recognition
sensitivity selection
and active
recognition
Training sensitivity
selection and active
training
Stand Alone mode
output ports 1-7
Stand Alone mode
output ports 8
Connection
Connect to microphone, other microphone
connection to GND
If powered mic, being used, NC, otherwise
connect JP2-1
GND
I/O
I
I
VCC
-
Connect to 8–32Ohm speaker. Provides
approximately 0.15 Watts of audio power
into 32-Ohms.
Connect to 8–32Ohm speaker. Provides
approximately 0.15 Watts of audio power
into 32-Ohms.
High-impedance (22kOhm) analog audio
output. Must be powered amplified to drive
a speaker, and should be low-pass filtered
with a corner frequency around 20KHz.
Better speech quality than the PWM output.
Recommended for applications requiring
either louder volume or better speech
quality.
To start recognition, pull
To erase all
the –RECOG line to GND
for at least 100ms. To erase recorded
words, pull
all recorded words, pull
both the –
both the –TRAIN and –
TRAIN and –
RECOG pins to GND for
RECOG pins
at least 100ms.
To start training, pull the – to GND for at
least 100ms.
TRAIN line to GND for at
least 100ms.
Connect to user application’s control lines.
O
O
Connect to user application’s control lines.
O
O
O
I
I
In this project, -TRAIN open circuit mode is used for pin configuration. This is relaxed
training mode – easier to train, accepts more similar sounding words (i.e., fewer
rejections).
11
Table 2: 15 Words and Corresponding Pinout
Since the module can recognize 15
words, but only has 8 output pins, word
9 through 15 are represented in binary
form, as show in the Table 2 on the right.
Software decoding method instead of
hardware decoding circuits is applied to
get word 9 and word 15.
Word 1
Word 2
Word 3
Word 4
Word 5
Word 6
Word 7
Word 8
Word 9
Word 10
Word 11
Word 12
Word 13
Word 14
Word 15
Output 1
Output 2
Output 3
Output 4
Output 5
Output 6
Output 7
Output 8
Output 8 and Output 1
Output 8 and Output 2
Output 8 and Output 3
Output 8 and Output 4
Output 8 and Output 5
Output 8 and Output 6
Output 8 and Output 7
2.1.3 SpeakJet Speech Synthesis Chip
The SpeakJet is a completely self contained, single chip voice and complex sound
synthesizer as shown in Figure 7. SpeakJet uses Mathematical Sound Architecture (MSA)
technology which controls an internal five channel sound synthesizer to generate on-thefly, unlimited vocabulary speech synthesis and complex sounds without the use of analog
or digitally recorded samples. The SpeakJet has a built in library of 72 speech elements
(allophones), 43 sound effects, and 12 DTMF Touch Tones. Through the selection of
these MSA components and in combination with the control of the pitch, rate, bend, and
volume parameters, the user has the ability to produce unlimited phrases and sound
effects, with thousands of variations, at any time.
The SpeakJet can be controlled simultaneously by logic changes on any one of its eight
Event Input lines, and/or by a Serial Data line from a CPU (such as Z8 or PC) allowing
for both CPU-Controlled and Stand-Alone operations. Other features include an internal
64 byte input buffer, internal programmable EEPROM, three programmable outputs, and
direct user access to the internal five channel sound synthesizer.
Figure 7: SpeakJet Chip (Left) and Pinout (Right)
12
The SpeakJet pin details and electrical specifications are shown in Figure 8.
Figure 8: SpeakJet Pin Details and Electrical Specifications
In the final project, Event Input E0-E7 is not used and must be connected to GND. The
Serial Input, RCX, is used to communicate between SpeakJet and external devices such
as Z8. PA5-TXD0 on Z8 board is connected to RCX. Voice Output, VOut, modulates the
SpeakJet’s voice on a square wave carrier of 32khz.
2.1.4 DS1307 Real-Time Clock (RTC) Module
This is a custom-designed module for the DS1307 Real Time Clock as shown in Figure 9.
The module comes fully assembled and pre-programmed with the current time. The
included Lithium coin cell battery (CR1225 41mAh) will run the module for a minimum
of 9 years (17 years typical) without external 5V power. The DS1307 is accessed via the
I2C protocol. It provides seconds, minutes, hours, AM/PM, day, month, date, year, leap
year compensation, accurate calendar up to year 2100, 1Hz output pin, and 56 Bytes of
non-volatile memory available to user.
Figure 9: RTC Module (Left) and RTC Module Schematic (Right)
13
VBAT is a battery input for any standard 3V lithium cell or other energy source. Vcc is the
+5V input. When 5V is applied within normal limits, the device is fully accessible and
data can be read and written. When a 3V battery is connected to the device and Vcc is
below 1.25xVBAT, reads and writes are inhibited. However, the timekeeping function
continues unaffected by the lower input voltage. SCL (Serial Clock Input) is used to
synchronize data movement on the serial interface. SDA (Serial Data Input/Output) is the
input/output pin for 2-wire serial interface. The SDA pin is open drain. SQW is used for
square wave output.
The DS1307 Serial RTC is a low-power, full binary-coded decimal (BCD) clock/calendar
plus 56 bytes of NV SRAM. Address and data are transferred serially via 2-wire, bidirectional I2C bus. DS1307 address map and timekeeper registers are shown in Figure 10.
Figure 10: DS1307 Address Map and Timekeeper Registers
2.1.5 MAX6610 Temperature Sensor
The MAX6610 is precise, low-power analog temperature sensors combined with a
precision voltage reference as shown in Figure 11. An 8-bit ADC’s LSB is equal to 1°C,
while a 10-bit ADC’s LSB corresponds to 0.25°C. The MAX6610 operates from 3.0V to
5.5V and has a 2.560V reference output. Power-supply current is less than 150µA. The
MAX6610is available in a 6-pin SOT23 package and operate from -40°C to +125°C.
Figure 11: MAX6610 Temperature Sensor Configuration
14
2.2.6 Parts List
Table 3 contains the part numbers used for this project.
Part Name
Z8 Encore! Development Kit
VoiceDirect Board
SpeakJet
Switch
SparkFun Real-time Clock Module
MAX6610 Analog Temperature Sensor
Microphone
Speakers
Breadboard
Wires
Quantity
1
1
1
2
1
1
1
2+
1
many
Description
Development package
Speech recognition board
Speech synthesis chip
Control speech recognition and Training
Real-time clock and calendar
Analog temperature sensor
For speech recognition
For speech recognition and synthesis output
For prototyping design and test
For connection
2.2 Software Modules
As shown in Figure 3, all hardware control and application software modules listed in the
block diagram of the software design are written by myself in C based on Z8 SDK. All of
them are compiled on Zilong XTools ZDS II – Z8 Encore! Family 4.9.6 (build 05110402).
ZDS II runs on the Windows XP platform. In addition, PhaseALator is used for some
basic English words or phrases editing for SpeakJet Allophone code creation. The
detailed software module development will be discussed in Section 3 Implementation &
Construction.
15
3 Implementation & Construction
The hardware construction is exactly followed by the schematic for the design in Figure 2.
Figure 12 shows the final Z8 Speaks system.
Speaker
Real-time Clock Module
Z8 Board
MAX6610
Temp. Sensor
VoiceDirect
SpeakJet
Speaker
Microphone
Switch for training
Switch for recognition
Figure 12: The Final Hardware Connection for the Z8 Speaks System
16
During the system development, for each major part or module needs to be tested on both
hardware and software side, respectively. In the following sections, the main modules
including hardware test and software development will be discussed in details. These
main modules are speech recognition, speech synthesis, real-time clock reading and
setting, temperature driver and reading, and phrase allophone editing. All other modules,
such as LED display, timer control, in the software block diagram as shown in Figure 3
can be implemented as easily as in the basic real-time embedded system. Their details
will not be given in this report. At the end of this section, the system integration and flow
chart will be described.
3.1 Speech Recognition
VOICEDIRECT board is used for speech recognition in this project. Firstly, hardware
structure needs to be tested and verified. Then the driver for Z8 needs to be written for
communication between VOICEDIRECT and Z8 microcontroller.
3.1.1 VOICEDIRECT Board Stand-Alone Test
VoiceDirect speech recognition board is connected in the stand-alone mode as shown
Figure 13 (a) and (b). The Z8 in the figure is only used to be a power supply and ground
connection. In (b), push left button at least 100ms, training will begin. VoiceDirect will
prompt “Say word x” (where x is number from 1 to 15 corresponding to the word to be
trained). After training, push right button, the VoiceDirect will prompt “Say a word”. If
the word matches the training records, the VoiceDirect will return to say the number of
the word record. Otherwise, it will return “Word not recognized”. Through this test,
hardware configuration and connection for the VoiceDirect board can be verified.
17
(a) The Schematic for VoiceDirect Stand-Alone Test
Switch for training
Switch for training
Switch for recognition
(b) The Hardware Connection for VoiceDirect Stand-AloneTest
Figure 13: Stand-Alone Test for VoiceDirect Hardware
18
3.1.2 VOICEDIRECT Driver Development for Z8
The VoiceDirect can output the signals of OUT1-OUT8 to communicate with Z8
microcontroller through the connections to Z8 GPIOs. I use PF0-PF7 to connect
VoiceDirect OUT1-OUT8 as shown in Figure 14.
(a) The Schematic for VoiceDirect Communication with Z8
(b) Hardware Connection for VoiceDirect Communication with Z8
Figure 14: Hardware Connection for VoiceDirect Communication with Z8
19
The following driver code snippets show how Z8 communicate with VoiceDirect. The
total 15 words are represented in binary form as shown in Table 2 and decoded as follows.
#define VOICE_COMMAND_1
#define VOICE_COMMAND_2
#define VOICE_COMMAND_3
#define VOICE_COMMAND_4
#define VOICE_COMMAND_5
#define VOICE_COMMAND_6
#define VOICE_COMMAND_7
#define VOICE_COMMAND_8
#define VOICE_COMMAND_9
#define VOICE_COMMAND_10
#define VOICE_COMMAND_11
#define VOICE_COMMAND_12
#define VOICE_COMMAND_13
#define VOICE_COMMAND_14
#define VOICE_COMMAND_15
0x01
0x02
0x04
0x08
0x10
0x20
0x40
0x80
0x81
0x82
0x84
0x88
0x90
0xA0
0xC0
// Initialize the ports for Voice Direct speech recognition board
void init_VoiceDirect(void)
{
// initialize Port F(0-7)
PFADDR = 0x02;
PFCTL &= 0x00;
// no alterate function
PFADDR = 0x01;
// data direction
PFCTL &= 0x00;
// clear
PFCTL |= 0xFF;
// input PF(0-7)
}
// Interpret the word code and take corresponding action
switch(PFIN) {
case VOICE_COMMAND_1: action1(); break;
case VOICE_COMMAND_2: action2(); break;
...
case VOICE_COMMAND_15: action15(); break;
default: break;
}
3.2 Speech Synthesis
SpeakJet chip is used for speech synthesis in this project. Firstly, hardware structure
needs to be tested and verified. Then the driver for Z8 needs to be written for
communication between SpeakJet and Z8 microcontroller.
3.2.1 SpeakJet Chip Stand-Alone Test
SpeakJet speech synthesis chip is connected in the demo (i.e. stand-alone) mode as
shown Figure 15 (a) and (b). The Z8 in the figure is only used to be a power supply and
ground connection. In such a mode, if all hardware works well, SpeakJet will play the
demo sounds.
20
(a) The Schematic
(b) The Hardware Connection
Figure 15: Hardware Test for SpeakJet Demo Mode
3.2.2 SpeakJet Driver Development for Z8
SpeakJet provides Serial Input and Event Inputs to communicate with Z8 microcontroller
through Z8 GPIOs. In Figure 16, Z8’s PG0 and PG1 connect to SpeakJet D0 and D1.
CTS0 (PA3 alternate function) connects to D2. Serial data line (TXD0) in Z8 board is
used to connect the RCX in SpeakJet. In this project, any Event Inputs are not used so all
of them connect to GND.
Figure 16: The Schematic for SpeakJet Communication with Z8
21
The SpeakJet can receive the commands that can be any of 256 commands listed in the
Table D & E in the attachment A, SpeakJet User Manual. There are 7 operational groups
of commands, SCP, Allophones, Sound Effects, DTMF, Pauses, Levels, and Controls. In
this project, SCP commands are not be used. Thus, each command received is buffered
into a 64-byte input buffer and executed by the MSA in FIFO.
Serial Data is the main command of communicating with the SpeakJet to execute
commands or create voices and sounds. The SpeakJet serial configuration is fixed at 8bits,
No-parity, and 1 stop bit and non-inverted. The SpeakJet can be configured to accept
Baud rates from 2400 to 19200. The factory default setting is 9600 baud. In this project,
default setting is used.
Therefore, the key for SpeakJet Z8 driver development is make serial data line TXD0
work. The driver code snippets are as follows,
#define FREQ 18432000
// 18.432MHz
#define BAUD1 9600
// 9.6K baud for UART0
#define BRG1 FREQ/((unsigned long)BAUD1*16)
#define UART_TXD_EN
0x80
// initialize SpeakJet speech synthesis chip
// Initialize the ports
void init_SpeakJet(void)
{
// initialize Port PA3 and PA5
PAADDR = 0x02;
PACTL |= 0x28;
// alterate function: PA3-CTS0, PA5-TXD0
PAADDR = 0x01;
// data direction
PACTL &= 0xD7;
// ouput:PA3-CTS0, PA5-TXD0
U0BRH = (char)(BRG1 >> 8);
U0BRL = (char)(BRG1 & 0xff);
U0CTL0 = UART_TXD_EN;
// Transmit enable, No Parity, 1 Stop
}
// Input the speech string into SpeakJet through serial data line
void Speak(char *speech)
{
puts(speech);
}
In this project, I directly use Allophones to edit any speech. For example, the following
allophone code string represents the English speech “Welcome to Z8 Speaks”.
unsigned char s_welcome[LEN_WELCOME] = {252, 252, 147, 159, 194, 134, 140, 8, 191, 162, 8, 167,
128, 128, 154, 4, 191, 187, 198, 8, 128, 196, 187, 1}; // “Welcome to Z8 Speaks”
22
3.3 Real-time Clock Reading/Setting
SparkFun Real-time clock module is used for read/set the time or calendar in this project.
RTC hardware specifications are described in Section 2.1.4. The schematics for SparkFun
DS1307 RTC module is shown in Figure 9. The driver for communication between RTC
and Z8 microcontroller contains the I2C protocol, data transfer, and time/calendar reading
or setting. These will be described in detail in this section.
3.3.1 I2C Protocol Control
I2C protocol is used for this module to communicate with Z8 microcontroller board. The
following functions are used for I2C communication. Note that DS1307 operates in the
regular mode (100KHz) only.
//Intialize I2C Interface
void init_I2C(void)
{
// BRG = 18432KHz/(4*100KHz) = 46 = 0x2E [mode = 100kHz for DS1307)
I2CBRH = 0x00;
// BRG High
I2CBRL = 0x2E;
// BRG Low
PAADDR = 0x02;
PACTL |= 0xC0;
I2CCTL = I2C_ENABLE;
// alterate function: PA6-SCL, PA7-SDA for I2C port
}
void I2C_start (void) {
I2CCTL |= SEND_START;
// I2C Start bit
}
void I2C_stop (void) {
I2CCTL |= SEND_STOP;
// I2C Stop bit
while ((I2CCTL & SEND_STOP) == SEND_STOP)
;
}
void I2C_write_byte (unsigned char data) {
I2CDATA = data;
// Write I2C data
}
void I2C_Transmit_Data_Empty (void) {
while ((I2CSTAT & TRANSMIT_DATA_REG_EMPTY) == 0x00) // Wait for Transmit
;
// Buffer Empty
}
unsigned char I2C_read_byte (void) {
unsigned char data;
data = I2CDATA;
//Read I2C data
return (data);
}
void I2C_Acknowledge (void) {
while ((I2CSTAT & RECEIVED_ACK) == RECEIVED_ACK) ;
// wait for ACK
}
unsigned char I2C_AckNack (void) {
unsigned char data;
while (1) {
// Wait for the last byte Transmit/Receive ACK/NACK
if ((I2CSTAT & RECEIVED_ACK) == RECEIVED_ACK)
23
return (ACK);
else if ((I2CSTAT & RECEIVED_NACK) == RECEIVED_NACK) {
I2CCTL |= FLUSH_TXD;
I2C_start ();
I2C_stop ();
data = I2C_read_byte ();
// Clear I2C Receive buffer
printf ("\n\n\rI2C Error - NACK Received");
return (NACK);
}}}
void I2C_Send_NACK (void) {
I2CCTL |= SEND_NACK;
// Send NOT Acknowledge
}
void I2C_Receive_Data_Full (void) {
while ((I2CSTAT & RECEIVE_DATA_REG_FULL) == 0x00) //Wait for Receive Buffer Full
;
}
3.3.2 Data Transfer
The DS1307 device address is 7-bit, 1101000. The last bit is for direction, 0 for write and
1 for read. The definition of the device identification code in the program is as follows,
#define EEPROM_Read_Address 0xD1
#define EEPROM_Write_Address 0xD0
//I2C EEPROM Read Address
//I2C EEPROM Write Address
The data transfer between Z8 and RTC is followed by 2-wire serial data bus. That is, the
bus is controlled by Z8 microcontroller that generates the serial clock (SCL), controls the
bus access, and generates the START and STOP conditions. The DS1307 operates as a
slave device on the serial bus. Access is obtained by implementing a START condition
and providing a device identification code followed by a register address. Subsequent
registers can be accessed sequentially until STOP condition is executed. Each receiving
device, when addressed, is obliged to generate an acknowledge after reception of each
byte. The DS1307 data write and read are shown in Figure 17.
(a) Data Read
(b) Data Write
Figure 17: DS1307 Data Transfer
24
The following code snippets show how the data transfer works (i.e. read/write) based on
I2C protocol.
unsigned char ReadEEPROM(unsigned char addrL)
{
unsigned char data;
I2C_start ();
I2C_write_byte (EEPROM_Write_Address);
I2C_Transmit_Data_Empty ();
I2C_Transmit_Data_Empty ();
I2C_write_byte (addrL);
if (!I2C_AckNack())
return;
I2C_Transmit_Data_Empty ();
I2C_start ();
if (!I2C_AckNack())
return;
I2C_Send_NACK ();
I2C_write_byte (EEPROM_Read_Address);
I2C_Receive_Data_Full ();
data = I2C_read_byte ();
I2C_stop ();
// I2C Start
// EEPROM Address; Write
// wait for transmit buffer empty
// wait for transmit buffer empty
// LSM EEPROM Address
// wait for ACK/NACK from EEPROM
// wait for transmit buffer empty
// I2C Start
// wait for ACK/NACK from EEPROM
// send NACK to EEPROM
// EEPROM Address; Read
// wait for receive data
// EEPROM Read data
// I2C Stop
return data;
}
void WriteEEPROM(unsigned char addrL, unsigned char data)
{
I2C_start ();
// I2C Start
I2C_write_byte (EEPROM_Write_Address);
// I2C EEPROM Address; Write
I2C_Transmit_Data_Empty ();
I2C_write_byte (addrL);
if (!I2C_AckNack())
return;
I2C_Transmit_Data_Empty ();
I2C_write_byte (data);
if (!I2C_AckNack())
return;
I2C_Transmit_Data_Empty ();
I2C_stop ();
if (!I2C_AckNack())
return;
// wait for transmit buffer empty
// LSB EEPROM Address
// wait for ACK/NACK from EEPROM
// wait for transmit buffer empty
// EEPROM Write data
// wait for ACK/NACK from EEPROM
// wait for transmit buffer empty
// I2C Stop
// wait for ACK/NACK from EEPROM
}
3.3.3 Time/Calendar Reading/Setting
DS1307 Address Map and Timekeeper Registers are shown in Figure 10. The following
function shows how to read the time and calendar from RTC.
#define SECOND
#define MINUTE
#define HOUR
0
1
2
25
#define DAY
#define DATE
#define MONTH
#define YEAR
3
4
5
6
// Read data from DS1307 RTC through I2C
unsigned char Read_RTC(int mode)
{
unsigned char data, tmp, tms;
data = ReadEEPROM(mode);
if(mode == HOUR) {
if(data&0x20)
AM = 1;
// AM
else AM =0;
// PM
tmp = data & 0x10;
tmp >>= 4;
tms = data & 0x0F;
data = tms + tmp*10;
}
else if(mode == DAY) {
data &= 0x07;
}
else {
tmp = data & 0xF0;
tmp >>= 4;
tms = data & 0x0F;
data = tms + tmp*10;
}
return data;
}
Likewise, using function WriteEEPROM(mode, address) can set the time or calendar.
3.4 Temperature Driver and Reading
MAX6610 analog temperature sensor is used for temperature reading in the final project.
I directly use MAX6610 sensor on the Lab3 board for this project. MAX6610 hardware
specifications are described in Section 2.1.5. The MAX6610 configuration and
schematics for Z8 are shown in Figure 11 and Figure 2, respectively. The driver for
communication to Z8 microcontroller contains ADC initialization and calculation for the
temperature as shown in the following code functions. ALG0 – PB0 port is used for
temperature input to Z8.
// Initialize the ADC
void init_ADC(int mode)
{
PBADDR = 0x01;
PBCTL |= 0x01;
PBADDR = 0x02;
PBCTL |= 0x01;
ADCCTL = 0x90;
// set up port B to use as ADC input
// PB0
// Port B alternate function
// enable PB0 = ANA0 ADC INPUT 0
// ADC Control Register
// CEN =1, 0, VREF=0, CONT=1, ANAIN=0000
26
IRQ0ENH |= 0x01;
// Set Interrupt Highest Priority
IRQ0ENL |= 0x01;
SET_VECTOR(ADC, isr_adc);
//set the interupt vector
EI();
// CEN will be 0 at first complete conversion result 5129+24 system clock cycles
while(ADCCTL & 0x80)
; // Do nothing but wait for the end of the first A/D conversion
}
//Interrupt routine
//ADC interupt.
#pragma interrupt
void isr_adc(void)
{
adc_data = calADC();
}
unsigned int calADC(void)
{
unsigned char tmp;
unsigned int data;
tmp = ADCD_L;
tmp >>= 6;
tmp &= 0x03;
data = ADCD_H;
data <<= 2;
data &= 0x03FC;
data |= tmp;
// get temperature
// get low bits
// move to the right side
// clean up
// move to the left side
// clean up
// ADC value
return data;
}
3.5 Phrase Allophone Editing
To produce speech, a list of selected allophones is sent to the SpeakJet. As the SpeakJet is
vocalizing this list of allophones, MSA actively and continuously calculates all the sound
components of the allophones including the transitional sounds made between the
allophones, producing the same sounds that the human mouth does as it moves one
position to another position.
Selecting the appropriate combination of allophones and pauses can thusly create any
English word or phrase. Further tuning with the Rate, Pitch, Bend and Volume
parameters adds to the delivery of the phrase and can change the emotion in which the
phrase is perceived.
Stressing the Rate, Pitch, Bend and Volume parameters to levels outside the human range
can result in some interesting sounds that go way beyond what a normal human mouth
can produce. In addition, several other sounds effects, which are included in the MSA
Sound Component Database, of which, some use vocalization and some do not, can be
integrated into the phrases.
27
The result is a system that gives the user the ability to not only produce an unlimited
vocabulary, but also to produce slang, gibberish, moans, groans, yodels and other weird
vocalized sounds not normally included in a canned TTS system.
All allophones in this final project are edited directly by hands and PhaseALator software.
These allophones mainly contain temperature, time, calendar, songs, etc. Although
allophones are mainly used for English language, they can be used for simulation of some
foreign language pronunciation.
3.5.1 Temperature Allophone Editing
The temperature phrase contains two major parts. One is the numerical phrase. The other
is just “degree” word. The allophone code of the “degree” can be directly translated from
PhaseALator. However, numerical phrase should be composed with 10 basic digit
numbers (i.e, 0, 1,…, 9) and 9 basic high digit numbers like 10, 20,…, 90, 100, 1000,... In
this project, all the allophones for these basic phrase elements are first stored in the ROM.
When used, they will be copied into the RAM. The following function shows how to
compose the speech of the temperature degree.
void SpeakTemperature(float temperature)
{
int T, T1, T0, lc;
unsigned char tmp[20];
T = (int)(temperature + 0.5);
if(T > 0) {
if(T <= 20){
lc = copy_data_digitCode(speech, CODE_1, T, 0);
datacat_rom(speech, s_degree, lc, LEN_DEGREE);
}
else {
T1 = (int)(T/10);
T0 = (int)(T%10);
lc = copy_data_digitCode(speech, CODE_10, T1-2, 0);
copy_data_digitCode(tmp, CODE_1, T0, 0);
datacat(speech, tmp, lc, lens[T0]);
datacat_rom(speech, s_degree, lc+lens[T0], LEN_DEGREE);
}
}
puts(speech); // Speak the temperature
}
3.5.2 Time/Calendar Allophone Editing
The time phrase contains two major parts. One is the numerical phrase. The other is just
“AM” or “PM” word. Likewise, the allophone code of the “AM” or “PM” can be directly
translated from PhaseALator. Numerical phrases are composed with 10 basic digit
numbers and 9 basic high digit numbers from 10 to 60. all the allophones for these basic
phrase elements are first stored in the ROM. When used, they will be copied into the
RAM. The following function shows how to compose the speech of the time.
void SpeakTime(unsigned char hour, unsigned min, int am)
{
28
unsigned char tmp[MAX_CODE_LEN], lc, T1, T0;
lc = copy_data_digitCode(speech, CODE_1, hour, 0);
if(min <= 20) {
lc = copy_data_digitCode(speech, CODE_1, min, lc);
}
else {
T1 = (int)(min/10);
T0 = (int)(min%10);
lc = copy_data_digitCode(speech, CODE_10, T1-2, lc);
copy_data_digitCode(tmp, CODE_1, T0, 0);
datacat(speech, tmp, lc, lens[T0]);
lc += lens[T0];
}
if(am)
datacat_rom(speech, s_AM, lc, LEN_AM);
else
datacat_rom(speech, s_PM, lc, LEN_PM);
puts(speech);
// Speak the time
}
The calendar phrases contains more elements, including day, date, month, and year.
Likewise, the calendar speech code strings can be composed by these elements using
similar methods.
3.6 System Integration and Main Application Flow Chart
The system integration contains two aspects, hardware integration and software
integration. Hardware components are integrated as shown in Figure 2 and Figure 12.
Software modules are integrated into one main application based on the software design
block diagram in Figure 3. This application includes 12 sub-applications corresponding to
15 voice commands.
In this VoiceDirect board for the final project, since OUT5 PIN16 does not work, the
Command 5 and Command 13 cannot be executed. In my program, Command 4 is not
used either. Thus, the total 12 voice commands or questions are listed as follows,
1) Command 1: welcome
2) Command 2: temperature
3) Command 3: time
4) Command 6: date
5) Command 7: name
6) Command 8: day
7) Command 9: math
8) Command 10: sing
9) Command 11: music
10) Command 12: goodbye
11) Command 14: set the time
12) Command 15: Ni Hao
-- "Hello!"
-- "Can you tell me the temperature?"
-- "What time is it now?"
-- "Can you tell me the date?"
-- "What's your name?"
-- "What day is today?"
-- "7 plus 8"
-- "Can you sing?"
-- "pa, pa" clap the hands
-- "Thank you! Bye-bye!"
-- "Reset the time"
-- "Ni Hao" Hello in Chinese
The flow chart of the main application for Z8 Speaks final project is shown in Figure 18
as follows.
29
START
Initialization
N
Train/ Record
Y
Train/Record Voice Commands
Speak Commands
Speech Recognition
Command 1
Welcome
Command 2
Temperature
Command 3
Time
Command 4
(Not Used)
Command 5
(Not Used)
Command 6
Date
Command 7
Name
Command 8
Day
Command 9
Math
Command 10
Sing
Command 11
Music
Command 12
Goodbye
Command 13
(Not Used)
Command 14
Set the Time
Command 15
Ni Hao
N
Y
Change Voice
Speak
N
Stop
Display
Y
END
Figure 18: The Flow Chart of Z8 Speak Main Application
30
4 Conclusions
The Z8 Speaks System is a hardware-software mix project. It integrates Z8
microcontroller, speech recognition, speech synthesis, real-time clock, temperature
sensing, LED display, timer control, and other hardware and software module
technologies.
The overall work for the final project works very well as planned in the project proposal.
This final work makes the user be able to interact with Z8 Speaks (i.e., a talking robot)
over the speech communication. The final project meets very well all the proposed
requirements, such as read the temperature from ADC temperature sensor, read the RealTime Clock (RTC) using I2C bus, control the speech recognition board using GPIO and
hardware switches, and control speech synthesis through serial data line. In addition,
more voice operations and speech controls are added into the final project beyond the
proposal.
Just as described in Section 3, Implementation and Construction, each step is an
important design decision I made for this project. After each step passes test and
verification, the system can be integrated correctly. In summary, these decisions are as
follows,
• Test and verify speech recognition
VOICEDIRECT board stand-alone test
OICEDIRECT driver development for Z8
• Test and verify speech synthesis
SpeakJet chip stand-alone test
SpeakJet driver development for Z8
• Test and verify SparkFun real-time clock module
Driver development through I2C
Time/Calendar management
• MAX6610 Temperature Driver development through ADC
• Phrase allophone editing
• System integration and main application development
During my implementation, strictly following each step not only made all of them work
very well but also reduce mistake or errors further accelerated the whole project
development. This further proves that decision making is the key for the development.
This is also one key point I learnt from this project. In addition, what else I have learnt
from this project is as follow,
• Hardware design and Z8 programming
• Speech recognition board use and development
• Speech synthesis chip use and development
• I2C control of Real-time Clock
• ADC control of temperature sensor
31
5 Attachments List
The list of the attachments for this final project is as follows,
Data Sheets
1) Sensory VoiceDirect Speech Recognition Kit (hardcopy)
2) SpeakJet User’s Manual (PDF)
3) DS1307 64x8 Serial Real-Time Clock (PDF)
4) SparkFun RTC Module Schematic (PDF)
5) MAX6610/6611 Temperature Sensors and Voltage References (PDF)
Source code
1) Speaks ZDSPROJ file
2) main.c, main.h
3) gpio.c, gpio.h
4) adc.c, adc.h
5) timer.c, timer.h
6) eeprom.c, eeprom.h
7) I2C.c, I2C.h
8) rtc.c, rtc.h
9) speakjet.c, speakjet.h
10) voicedirect.c, voicedirect.h
11) zsldevinit.asm
Demonstration Video (13MB)
Liao_z8Speaks_320x240.mpg(13MB)
Reports
Project proposal (PDF)
Project Status Report (PDF)
Final Project Report – Z8 Speaks (PDF)
Project Summary brochure (PDF)
32
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement