Dialogic 6.2 Manuale Utente

Making Two-Channel Call Transfers (Tromboning)

November 2009

194

The application created a new connection from the output of network 

timeslot 0 to the input of network timeslot 1 and another new 

connection from the output of network timeslot 1 to the input of 

network timeslot 0 (see 

). 

Using this model, Channel 0 can record the data that is being 

transmitted from the network timeslot 0 and Channel 1 can record 

the data that is being transmitted from network timeslot 1.

Because of firmware requirements, when disconnecting resources, you 
have to invert the source and destination resources. For example, if you 
made a transmit connection from the output slot of Channel 0 to the input 
slot of network timeslot 0, where Channel 0 is the source, then you must 
disconnect from the input slot of network timeslot 0 to the output slot of 
Channel 0.

After the application establishes a two-channel call transfer, it can 

then perform voice playback and voice recognition. In addition to 

these voice actions, the application must provide secondary 

functionality. For example, when performing voice recognition 

during a two-channel call transfer, the application must configure 

echo cancellation in the channel correctly. Also, when performing 

voice playback to a caller, the application must place the other caller 

on hold to avoid providing the original caller with two source inputs.

When a caller sends voice data over the telephone network, the 

network reflects some of that data back as input data. The reflected 

data is then mixed in with valid input voice data from the remote 

end. The channel that is connected to the telephone network must be 

able to remove the reflected data from the valid input data. This 

technique is called echo cancellation. 

When a channel performs echo cancellation, it analyzes the voice 

data that is being sent out into the telephone network. When this 

voice data reappears as input voice data (the echo), the channel 

removes the data from the input data stream. What is left is the 

actual voice data from the remote end.