Packet Time

The interval in milliseconds, for transmitting each audio packet.

G729

G.729 codec is used for audio data compression for both H.323 and SIP connections. It supports G.729 Annexes A and B.

Reduced Complexity Encoder

The G.729A reduced complexity encoding algorithm lowers the performance cost of G.729 transcoding. This setting affects only the encoding of G.729 audio sent by Experience Portal. The audio quality is reduced slightly when you enable this option.

Experience Portal continues to receive and decode G.729 and G.729A audio data, regardless of the option selected in this field.

Discontinuous Transmission

The G.729B discontinuous transmission algorithm allows Experience Portal to access and process a far end media offer with G.729B. The Annexe B specification further reduces network bandwidth as it sends only the audio packets that contain speech data (packets that contain silence are not transmitted).

Threshold

The options are:

• Voice — The Voice Threshold parameter is used by the call classification engine in determining whether a given frame of audio data should be interpreted as voice energy. Lower values are more inclusive and tends toward marking even white background noise as voice. Higher values will reject more noise, picking out only audio frames that are very clearly voice sounds. The range is 0.0-1.0 and the default value is 0.50

• Tone — The Tone Threshold parameter is used conjunctively along with Periodicity Threshold parameter by the call classification engine to determine if a given frame of audio data contains a pure telephony tone. Lower values for either or both will allow the call classification engine to accept more distorted signals as valid telephony tones, but makes certain voice sounds more likely to be detected as pure tones, for example, Talk Off. The range is 0.0-1.0 and the default value is 0.95.

• Periodicity — The Periodicity Threshold parameter is used conjunctively with Tone Threshold by the call classification engine to determine if a given frame of audio data contains a pure telephony tone. Lower values for either or both will allow the call classification engine to accept more distorted signals as valid telephony tones, but makes certain voice sounds more likely to be detected as pure tones, for example, Talk Off. The range is 0.0-1.0 and the default is 0.97.

• Ring Count — Ring Count Threshold detects the number of ring back cycles . The purpose is to let the call progress engine to use different timers when a call is answered “quickly”, as determined by the ring count, tailoring the parameters appropriately under the assumption that calls answered after a couple of rings are more likely to be a live person. The cycles range from 0-8 and the default value is 4.

Cut Through

The Cut Through time is the number of milliseconds of consecutive silence after some voice energy that must be heard to determine that a live speaker is done talking.

The following are the values of Cut Through time:

• Initial — The Initial values are loaded first when the call classification engine is initialized for any given call. The value ranges from 200-2000 milliseconds and the default is 1100 milliseconds.

• Short — The values for the Short timers are loaded when the call classification engine detects at least one ring back cycle . The value ranges from 200-2000 milliseonds and the default is 700 milliseconds.

• Long — The values for the Long timers are loaded if the call classification engine detects successive ring back cycles and the count equals or exceeds the Ring Count Threshold. The value ranges from 200–2000 milliseconds and the default is 1100 milliseconds.

Which value is used is determined by the number of ring back cycles heard by the call classification engine and the Ring Count Threshold. The purpose is to let the call progress engine to use different timers when a call is answered “quickly”, as determined by the ring count, tailoring the parameters appropriately under the assumption that calls answered after a couple of rings are more likely to be a live person.

Max Voice

The Max Voice time is the number of milliseconds of continuous voice energy except gaps of silence shorter than the Cut Through time that must be heard to determine that a greeting is a recorded message.

The following are the values of Max Voice time:

• Initial — The Initial value is loaded first when the call classification engine is initialized for any given call. The value ranges from 1100-4000 milliseconds and the default is 2500 milliseconds.

• Short — The values for the Short timers are loaded when the call classification engine detects at least one ring back cycle. The value ranges from 1100–4000 milliseconds and the default is 2500 milliseconds.

• Long — The values for the Long timers are loaded if the call classification engine detects successive ring back cycles and the count equals or exceeds the Ring Count Threshold. The value ranges from 1100–4000 milliseconds and the default is 2500 milliseconds.

Which value is used is determined by the number of ring back cycles heard by the call classification engine and the Ring Count Threshold. The purpose is to let the call progress engine to use different timers when a call is answered “quickly”, as determined by the ring count, tailoring the parameters appropriately under the assumption that calls answered after a couple of rings are more likely to be a live person.