 |
 |
 |
 |
 |
|
|
|
|
|
Glossary
application programming interface (API)
A set of standard software interrupts, calls, and data formats that application programs use to initiate contact with network services or other program-to-program communications.
A function that allows program execution to continue without waiting for a task to complete. To implement an asynchronous function, an application-defined event handler must be enabled to trap and process the completed event. See synchronous function.
automatic speech recognition (ASR)
A set of algorithms that processes speech utterances.
The act of a party beginning to speak while a prompt is being played. When the VAD detects significant energy in the voice channel, CSP can optionally terminate prompts playing on that channel. Thus the party on the other end of the line is said to have "barged in" on the prompt.
A block of memory or temporary storage device that holds data until it can be processed. It is used to compensate for the difference in the rate of the flow of information when transmitting data from one device to another.
comfort noise generation (CNG)
The ability to produce a background noise when there is no speeech on the telephone line.
The point at which the echo canceller processes enough data to be able to identify the echo component in the incoming signal and thereby reduce it to provide echo-cancelled data to the host.
A computer peripheral or component controlled through a software device driver. An Intel Dialogic voice and/or network interface expansion board is considered a physical board containing one or more logical board devices, where each channel or time slot on the board is a device.
Dialogic mediastream processing architecture. It is open, layered, and flexible, encompassing hardware as well as software components. A whole set of Dialogic products are built on DM3 architecture.
A software module that provides a defined interface between a program and the hardware.
The component of an outgoing signal (that is, the play prompt) reflected in the incoming signal. The echo occurs when the signal passes through an analog device or other interface somewhere in the circuit.
The incoming signal whose echo component has been significantly reduced by the echo canceller.
A technique used to significantly reduce traces of an outgoing prompt in the incoming signal. These traces are referred to as echo. The echo canceller is the component in CSP responsible for performing echo cancellation.
A set of program instructions that are resident (usually in EPROM) on an expansion board.
In this configuration, the resource devices (voice/fax) and network interface devices are permanently coupled together in a fixed configuration. Only the network interface timeslot device has access to the CT Bus.
In this configuration, the resource devices (voice/fax) and network interface devices are independent, which allows exporting and sharing of the resources. All resources have access to the CT Bus.
incoming signal or incoming speech signal
The speech uttered by the caller, or the DTMF tone entered by the caller. Also known as the echo-carrying signal. This signal contains an echo component only if an outgoing prompt is played while the incoming signal is generated.
The lag time experienced as a result of audio energy traveling over the telephone or data network from the sender to the receiver.
A collection of precompiled routines that a program can use. The routines, sometimes called modules, are stored in object format. Libraries are particularly useful for storing frequently used routines because you do not need to explicitly link them to every program that uses them. The linker automatically looks in libraries for routines that it does not find elsewhere.
A process used to block or suppress the residual (echo-cancelled) signal, when there is no near end speech. This process can be used with comfort noise generation (CNG) to produce background noise. Background noise energy estimation is used to adjust the level of comfort noise generated. This allows the speaker to listen to the same level of background noise when the non-linear processor is switched on and off due to double-talk situations or near end speech. A typical usage of this feature is background noise used in dictation applications to let the user know that the application is working.
outgoing prompt or outgoing signal
The speech in a computer telephony application that is played to a caller. Also known as the echo-generating signal.
A circular buffer that stores the incoming speech signal and is used to reduce the problem of clipped speech. This data, which includes the incoming speech signal prior to the VAD trigger, is then sent to the host application for processing. This action ensures that minimal incoming data is lost due to VAD latency.
reference signal or echo-reference signal
The outgoing signal that is free of echo before it is passed to the network device. This signal is used by the echo canceller to characterize the echo to be removed from the incoming signal.
Class of functions that take one input parameter (a valid Dialogic device handle) and return generic information about the device. For instance, Standard Attribute functions return IRQ and error information for all device types. Standard Attribute function names are case-sensitive and must be in capital letters. Standard Attribute functions for all Dialogic devices are contained in the Dialogic SRL. See Standard Runtime Library.
A Dialogic downloadable signal- and call-processing firmware. Also refers to boards whose device family is not DM3.
A Dialogic software resource containing Event Management and Standard Attribute functions and data structures used by all Dialogic devices, but which return data unique to the device.
A function that blocks program execution until a value is returned by the device. Also called a blocking function. See asynchronous function.
Also called tail length or length. Refers to the number of milliseconds of echo that is eliminated from the incoming signal. The length of an echo canceller is sometimes given as "taps," where each tap is 125 microseconds long.
Time-division-multiplexed bus. A resource sharing bus such as the SCbus or CT Bus that allows information to be transmitted and received among resources over multiple data lines.
Also called voice energy detector. This detector identifies the presence of speech energy and determines when significant energy is detected in the voice channel. It notifies the host application that speech energy is detected.
|
|
|
|
|
Click here to contact Dialogic Customer Engineering
Copyright 2001, Intel Corporation