80-0206-R Datasheet, PDF (29/49 Pages) List of Unclassifed Manufacturers – Speech Recognition Processor

Data Sheet
RSC-4128
Because it is intended to “listen” continuously at very low power levels, the Audio Wakeup unit must detect each of
these events without any processor interaction. The processor configures and enables the unit under program
control before going into Idle mode. Audio Wakeup is not available in Sleep mode because the unit requires the
CLK2 signal. The detection signal from the Audio Wakeup unit can trigger a wakeup event, which starts the
processor and allows further audio processing. The processor inputs to the Audio Wakeup are an enable signal and
control signals to select for which sound to listen. See schematic 1-3 for details on this implementation.
Schematic 1-3
NOTES:
1. Optional. This capacitor MAY reduce noise coupled into the
mic input on a noisy PCB.
2. If used, this capacitor MUST be placed close to the
RSC-4128 AGND and MIC1IN pads.
3. Place close to MICIN1.
4. These capacitors may be used to reduce high frequency
response. The tradeoff is reduced amplitude.
5. For reducing system ESD susceptibility.
6. Typical value. Refer to Sensory Design Note 80-0259
"Selecting a Microphone" to determine proper value for your
application
7. SETTING RESISTOR AND CAPACITOR VALUES IN THE MICROPHONE
POWER LINE FILTER CIRCUIT:
The capacitor value should be between 220 and 30 ufd.
The resistor value should be between 300 and 2000 ohms.
Within these ranges;
Select the largest capacitor consistent with other
constraints (cost, size,etc.)
For this capacitor, select a resistor such that the RC
time constant is ~60msec.
MK1
1
2
MICROPHONE
* Refer to Sensory Design Note 80-0073 "Speech Recognition
Hardware Design" for information on proper microphone
selection and housing design, PCB layout, as well as test and
evaluation procedures to verify proper design and operation.
* Sensory offers a FREE Design Review Service. Detatils of
this service are also included in 80-0073.
NOTE 6
R2
NOTE 7
R1
1.8K
C5
220uF
NOTE 7
C8
.1
270
AVdd
R3
3300pF
NOTE 1
C9
NOTE 2
510
D1
1N4148
C7
.1
C10 D2
.01
1N4148
R4
510
1N4148
MK1
1
2
MICROPHONE
NOTE 6
R2
1.8K
C5
220uF
NOTE 7
NOTE 7
R1
C7
.1
270
AVdd
D1
R3
1N4148
RSC-4x
Example using one
microphone for normal
operation only
Px.n (Px.n is any available port
I/O pin)
C11
.01
NOTE 5
Recommended
C6
.1
NOTE 3
C1
.1
C4
.1
Vdd
510
C8 NOTE 1
3300pF
NOTE 2
C9 D2
.01
1N4148
C2
2.2uF
C3
1uF
NOTE 5
Recommended
RSC-4x
Example using one microphone for
both Audio Wakeup and normal
operation
Px.n (Px.n is any available port
I/O pin)
NOTE 4
C6
.047
C1
.1
NOTE 3
C2
2.2uF
D3
C4
.1
C3
1uF
Vdd
D4
1N4148
AVdd
Vdd
BT1
3V
The RSC-4128 FluentChip™ library contains routines for detecting each of the four audio events listed above.
These routines also manage powerdown appropriately. See the “FluentChip™ Technology Library Manual” for
reference code to invoke these routines.
Microphones
A single electret microphone may be used both for the analog front-end input (for recognition purposes) and as the
sound source for the Audio Wakeup unit. The current consumption and frequency response requirements are
different for the two uses, so two microphone input pads are provided: MICIN1 for the normal recognition input to
the analog front-end, and MICIN2 for the Audio Wakeup analog front end. A common microphone ground is used
for both the normal recognition analog front-end and the Audio Wakeup analog front end.
During normal recognition and Audio Wakeup operation, the microphone would typically be powered from a source
with an impedance in the range of 1-3 Kohms. If both the normal recognition and Audio Wakeup front ends are
used, they must be isolated from each other by capacitors and may share one microphone and microphone bias
circuit. The switching of the microphone input source is under program control. See schematic 1-3 for details on
this implementation.
The recommended value for the microphone filter capacitor (labeled “C5” in Schematic 1-3) is in the range of 33uF-
220uF. Using a capacitor at the upper end of this range will reduce low frequency noise. Low frequency noise on
the microphone input typically won’t affect recognition, but could affect the quality of speech playback when using
Record and Playback technology in an application. (see the “FluentChip™ Technology Library Manual” for more
information on Record and Playback) Typical low frequency noise sources include 60 Hz hum, “motor boating” or
29
P/N 80-0206-R
© 2006 Sensory Inc.