| 64kbps |
64 kilobits per second data rate |
The rate of data transmission on a single channel of E1, T1, etc. Data storage at 64kbps is
usually a verbatim copy of the signal on the stream, although it is possible to
convert from A-law to Mu
-law, remaining at 64kbps but modifying the binary data. |
| 32kbps |
32 kilobits per second data rate |
OKI ADPCM and G.721/G.726 ADPCM effectively use 4 bit code words for each signal
sample. This uses half the amount of data as A-law
and mu -law to describe the speech signal.
|
| 24kbps |
24 kilobits per second data rate |
This data rate is either OKI ADPCM (4 bits per sample) sampled at
6kHz, OR G.726 in a mode using 3 bits per
sample. |
| 16kbps |
16 kilobits per second data rate |
Prosody uses a proprietary data
compression algorithm which encodes speech at 16kbps, a quarter of the data
rate of G.711. |
| 12kbps |
12 kilobits per second data rate |
Prosody uses a low quality data
compression algorithm which encodes speech at 12kbps. |
| 8kbps |
8 kilobits per second data rate |
Prosody supports a proprietary high quality CELP algorithm, which records speech using
1/8th of the data rate of G.711 |
| ADPCM |
Adaptive Differential PCM |
A variant of PCM, in which each code word
corresponds to the difference between the signal level and an internally
predicted value for the signal level. The scaling of the code word is adapted
based on signal history, as is the logic for making signal level
predictions. |
| A-law |
G.711 |
A form of PCM, whereby the code word maps
non-linearly to the signal value. A-law code words
are 8 bits wide. Used for the B-channelin ISDN. |
| algorithm |
|
In general, a set of rules, usually followed by a computer. A process
implemented in software, e.g. 16kbps data
compression |
| API |
Application Programmer Interface |
The set of function calls used by a programming language such as C, to
access and operate the hardware product. e.g. call_openin(), sm_play_wav_file
() |
| ASR |
Automatic Speech Recognition |
An algorithm running on a computer telephony platform such as Prosody which recognises spoken words. |
| Audiotex |
|
An automated system which transmits recorded speech over the
telephone, usually a Premium Rate
service |
| Barge-in |
|
In ASR, barge in refers to the user speaking
while the system is issuing an outgoing prompt. |
| Basic rate
(BRI) |
|
A digital telephony trunk carrying two 64kbps
B-channels and one 16kbps D-channel.
Also referred to as ISDN-2 |
| B-channel |
Bearer Channel |
In digital telephony, the channel which
carries the subscribers information usually speech, but possibly
modulated data (e.g. fax) or raw data (e.g. V.110) |
| BONDING |
Bandwidth on Demand |
A single telephone call that uses of a number of 64kbps B-channels to
provide bandwidth for a call that uses higher bit-rates. Without BONDING,
transfer of this data would require a number of separate calls to provide the
bandwidth. |
| Cadence |
|
The on / off sequence of a tone or set of tones. Different call-progress tones often share the same pitch of
tones, but have different cadences. |
| Call-Progress
tones |
|
Specifically, tones generated on a telephony channel to indicate the status of a telephone call,
e.g. Path Engaged, Ring Tone. Generally, the term is used for any tone which is
specified by pitch and cadence. |
| CELP |
Code-Excited Linear Prediction |
Whereas PCM and ADPCM encode each sample of speech using an integer
(together with linear prediction), CELP algorithms encode short sequences of
samples with a single symbol. The symbol indexes into a codebook of potential
sequences, which is known to the encoder and decoder . |
| CCITT |
Consultative Committee on International Telegraph and
Telephone |
The standards body for the telephony industry, Now replaced by the
ITU. |
| channel |
|
The carrier of a single (speech or data) signal. Can be simplex or duplex |
| Codec |
coder decoder |
A pair of algorithms for encoding and subsequently decoding a signal,
e.g. a speech signal at 8kbps |
| Conference |
|
Broadly, the addition of two or more speech signals onto one channel. In a teleconference, the signal transmitted to
each listener is the sum of the speech from the other parties. |
| Connected-Word
Recognition |
|
Automatic speech recognition (ASR) which
allows the speaker to say words without spaces inbetween. Connected word
recognisers tolerate co-articulation (e.g. "Two Eight" becomes "Tooweit")
|
| CRC |
Cyclic Redundancy Check |
A method for verifying the integrity of a block of stored or
transmitted data. |
| CTI |
Computer Telephony Integration |
|
| Cut-through |
|
A DTMF detector must be able to detect
DTMF tones in the presence of other interfering
signals. |
| D-channel |
Data channel |
In digital telephony, a channel which
carries signalling information for use by the telephone network. |
| Data
Compression |
|
An algorithm which allows a signal to be stored in a more compact form
than its original, "raw" format. |
| Device
Driver |
|
The software that provides communication with specialised hardware. A
software application accesses the device driver via its API |
| DSP |
Digital Signal Processing |
The manipulation or analysis of any signal (such as sound, pictures,
radio waves) by a digital computer. |
| DTMF |
Dual Tone, Multi-Frequency |
The set of dual tones generated by a typical "touch-tone" telephone
keypad. |
duplex, full
duplex, half duplex |
|
A duplex channel carries a signal in both directions between two
points. A full duplex channel can transmit in both directions simultaneously,
whereas a half-duplex channel alternates between directions. A full duplex
channel is effectively two channels, whereas a half-duplex channel alternates
direction. |
| Echo
Cancellation |
|
Elimination of side-tone from a return
signal. The transmitted signal is used as a reference. |
| File |
|
A lump of data stored in a computer. The data may be an executable
program, text, information readable by the computer, speech, etc. |
| Fixed
point |
|
Arithmetic which uses only integers. The integers usually represent
multiples of an independent scaling factor. The scaling must be handled
separately by extra logic. |
| Floating
point |
|
Numbers are stored as an integer and an explicit scaling factor.
Scaling is handled by the Floating Point Unit. |
| G.711 |
|
See A-law and m
-law |
| G.721 |
|
An ITU standard for compression of speech
signals to 32kbps, 48kbps or 24kbps. Now
obsolete, but replaced by G.726 |
| G.723.1 |
|
Otherwise known as TrueSpeech, a method for compressing speech signals
to a data rate of 7.6kbps. Used extensively in internet telephony (VoIP) |
| G.726 |
|
An ITU standard for compression of speech signals to 32kbps, 48kbps, 24kbps or 16kbps. Replaces
G.721 |
| Gain |
|
A multiplier applied to a signal. Usually refers to volume level.
Measured in dB. |
| Grunt
Detection |
|
Detection of any significant signal on a telephone line. Prosodys grunt detection estimates background
noise, and provides the application with the duration and average energy of the
"grunt" |
| GSM6.10 |
|
The method of speech data compression used in the GSM mobile telephone
system. |
| H.320 |
|
ITU standard for transmitting video on narrowband telephony
channels |
Hidden Markov Model
(HMM) |
|
A mathematical tool used to describe statistics of a changing, unknown
system based on observations of the effects of that system. In Automatic speech
recognition, an abstract statistical model of the "speech" is generated, based
on acoustic observations |
| ISA |
Industry Standard Architecture |
Similar to IBM-AT, a standard bus for connecting peripheral hardware
to a Personal Computer main board. |
| ISDN |
Integrated Services Digital Network |
A switched network providing end-to-end digital connectivity for
simultaneous transmission of voice and/or data over multiple multiplexed
communications channels and employing transmission and out-of-band signalling
protocols that conform to internationally defined standards. An ISDN line
consists of a collection of B-channels carrying
speech and/or data, and one or more D-channels
carrying signalling information. |
| ISDN-2 |
|
see Basic Rate |
| IVR |
Interactive Voice Response |
An IVR system communicates with a user using spoken prompts. The user
can control the system using either DTMF or pulse dialled digits, or a restricted set of spoken words
(ASR). |
Isolated Word
Recognition (IWR) |
|
Automatic speech recognition (ASR) which
requires the speaker to annunciate single words with spaces
inbetween |
| ITU |
|
International Telecommunications Union. Official body for ratifying
standards in telecommunications. |
| Mark/Space |
|
When describing a signal which periodically alternates between the on
state and the off state, the "mark" is the amount of time on, and the "space"
is the amount of time off. |
| MFLOPS |
Mega Floating-point Operations Per Second |
A measure of the power of a floating point processor. Since many
processors allow parallel floating point operations, this is often larger than
the number of MIPS (million instructions per second) |
| Modem |
Modulator -Demodulator |
A device or algorithm which converts data into a signal which can be
borne on a telephone line (either as an analogue signal or on the B-channel in digital telephony) and subsequently
converts this signal back to data. |
| mu - law /
µ - law |
G.711 |
An alternative to A-lawPCM, used for the B-channel
of T1 streams. |
| MVIP |
Multi Vendor Integrated Protocol |
An industry standard for connecting PC-based computer telephony
hardware. |
| OKI
ADPCM |
|
An efficient, low quality version of ADPCM
made popular by OKI, and now an industry standard. The scaling of the code
integer is adapted but there is no adaptive predictor. |
| Operating
System |
|
In a computer, the "system" or "executive" software, which oversees
all other "user" programs that perform specific tasks. |
| PCI |
Peripheral Component Interconnect |
An alternative to ISA bus for connecting
peripheral hardware to a Personal Computer main board. PCI allows much faster
data exchange than ISA and allows much of the device configuration to be
completed in software.ISA and allows much of
the device configuration to be done in software. |
| PCM |
Pulse Code Modulation |
A
method of describing a time-varying signal using a sequence discrete integers
as code words. Each integer code word describes the signal level at a point in
time. |
| Premium
Rate |
|
A telephone line or number with an increased tariff which provides
revenue for both the telecommunications provider (PTT) and the owner of the
telephone line. Information or entertainment services are often provided using
premium rate lines. |
| Primary
Rate |
|
A digital telephony trunk carrying 30 64kbps B-channels and two
64kbps D-channels |
| Prosody |
|
The powerful and versatile speech processing card from
Aculab. |
Pulse detection,
Pulse dialling, 10pps detection |
|
In the absence of DTMF capability,
telephones dial by repeatedly switching a current on the line. This is done at
a rate of approximately ten switches per second. The switching current does not
propagate beyond the local exchange and so is very difficult to detect
accurately by a remote system. This method of dialling is used by all rotary
dial telephones. |
PSTN
|
|
Public Services Telepone Network = the telephone network that the
public connect to! e.g. BT, NTL, Cable London. Your analogue or ISDN line
connects to such a network. It enables you to connect to anyone in the world!
via a telephone connection. |
| Q.25 |
|
An ITU specification for DTMF tones. |
| Recursive |
|
See "Recursive" |
| SCBUS |
|
Dialogic equivalent of MVIP |
| SHARC |
ADSP-2106x |
Analog Devices floating point Digital Signal Processor
chip |
| Side-tone |
|
The fraction of an outgoing signal which is reflected by an analogue
telephone. |
| Simplex |
|
A channel which allows transmission in
one direction only, for example terrestrial radio. All digital telephony
channels are simplex; a duplex connection is
provided by a pair of channels. |
| Stream |
|
A collection of 64kbps digital signals,
time-division multiplexed onto one physical transmission line. For example, a
single MVIP stream carries 32 x 64kbps signals, each of which is referred to by its
timeslot. |
| Switch |
|
In the context of computer telephony, switching is the connection of a
speech channel to a resource, or to the MVIP or SC bus. |
| T-1 |
|
A digital telephony trunk providing 24 64kbps
channels. In Robbed-bit T-1, the D-channel is buried within the B-channels, using least-significant bits. Otherwise,
T1 provides 23 B-channels and 1 D-channel. |
talk-off, Mitel,
Bellcore |
|
A DTMF detector can sometimes detect valid
tones in a speech signal, due to frequency components of the signal lying in
the same range as valid DTMF tones. This false
detection is "talk-off". There are two standard tapes of signals used to
measure talk-off performance, respectively by Mitel and Bellcore. |
| TTS |
Text to Speech |
A system which converts text to spoken language. |
| Timeslot |
|
The sub-division of a stream which
carries one channel of information. |
| Trans-coding |
|
Conversion of one recorded speech format to another, for example,
conversion from G.711 A-law to Prosody 16kbps. |
| V.29 |
|
Fax communication standard, for operation at 9600 bits per second,
with fall-back to 7200 bps |
| V.21 |
|
Standard for communication between modems or Faxes at
300bps |
| V.23 |
|
Standard for modem communication at 1200bps (with 600bps backward
channel) |
| V.110 |
|
|
| V.120 |
|
|
| VoIP |
Voice over Internet Protocol |
Any system which transmits real-time speech (usually duplex) as data over the internet. |
| VOX |
|
- Voice activated recording, or voice
activity detection;
- A file format consisting of raw
data, with no length or data format information, used by some speech
cards.
|
| WAV |
|
A standard format for storing audio files on a computer. The header on
a WAV file describes the format of the data that follows. The data itself can
be in virtually any format, as long as the format has a unique name, registered
with Microsoft. |