PSB2163T-V31TR Infineon Technologies, PSB2163T-V31TR Datasheet - Page 144

• Manufacturer Part Number: PSB2163T-V31TR
• Description: IC AUDIO RINGING CODEC 28-PDSO
• Manufacturer: Infineon Technologies
• Series: ARCOFI®r

Specifications of PSB2163T-V31TR

• Function: CODEC Filter
• Interface: IOM-2, PCI, Serial
• Number Of Circuits: 1
• Voltage - Supply: 5V
• Current - Supply: 900µA
• Power (watts): 1mW
• Operating Temperature: -25°C ~ 80°C
• Mounting Type: Surface Mount
• Package / Case: DSO-28
• Includes: Activation and Deactivation, B-Channel and D-Channel HDLC Controllers, D-Channel Priority Handler, Monitor Channel Handler
• Lead Free Status / RoHS Status: Request inventory verification / Request inventory verification
• Other names: PSB2163T-V31INTR; PSB2163T-V31TR

Speakerphone Implementation
between the average signal level and the instant signal level leads to speech
recognition, the influence of noise on the decision is cancelled, even if the noise level
exceeds the voice level.
Figure 5
Block Diagram of the Speech Detector
Advanced Signal Processing
Before the signal is applied to the low pass LP2, it is preprocessed by a logarithmic
amplifier, a second low pass LP1, and a peak detector.
The logarithmic amplifier compands the signal logarithmically, therefore the speech
detector got a high dynamic range. With the parameter LIM a threshold is defined. Input
signal levels that are below the threshold determined by LIM are not processed by the
speech detector. Usually LIM is programmed for a threshold a few dB above the
system’s noise floor. Note, that LIM is related to the maximum PCM value, causing for
example a threshold of -60dBm0 if programmed to LIM=
63.1 dBm0.
The main task of low pass filter LP1 is to filter the incoming signal in a way that main
spikes are eliminated. Due to the programmable time constant LP1 it is possible to
defuse as well high-energy syllables as noise edges.
The peak detector bridges the very short speech pauses during a monologue, so this
time constant has to be long. Furthermore, the speech bursts are stored, so that a sure
speech detection is guaranteed. But if no speech is recognized, the low pass LP2 must
be discharged very quickly to the averaged noise level. In addition, the noise edges are
Semiconductor Group
143