Data processing: speech signal processing – linguistics – language – Speech signal processing – For storage or transmission
Patent
1997-01-13
1999-08-17
Hudspeth, David R.
Data processing: speech signal processing, linguistics, language
Speech signal processing
For storage or transmission
379 8809, H04M 322
Patent
active
059407921
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to the testing of telecommunications systems.
2. Related Art
Modern telecommunications systems perform complex operations on the signals they handle in the process of transmitting the signals through the telecommunications network, for example digitisation and compression techniques. These operations have non-linear effects on the signal inputs and it is thus not possible to model the effects of the network by the simple additive effect of each component of the network. In particular, the effect of the network on speech is not easily derivable from studying its effect on a simple test signal such as a sine wave.
Various methods of deriving test signals which mimic generalised speech properties have been devised (see for example German Patent specification DE 3708002 (Telenorma), European patent specification EP0567439, and the present applicant's published International applications W094/00922 and WO 95/01011), but these must all presuppose certain conditions, and in particular they require the use of predetermined test signals. The use of live (real time) traffic as a test signal for these tests would be impossible. The test site (which may be many thousands of miles away from the signal source in the case of an intercontinental link) needs to have knowledge of the test signal, so that deviations from the test signal can be distinguished from the test signal itself. The use of prearranged test signals may also require cooperation between the operators of two or more networks. Moreover, any line carrying a voice-frequency test signal is not available for use by a revenue-earning call, as the revenue-earning call would interfere with the test, and the test signal would be audible to the makers of the revenue-earning call.
It is known to test lines carrying live data (as distinct from speech), but this is a relatively simple problem because the information content of the signal consists of only a limited range of signals (e.g. DTMF tones, or binary digits), and it is relatively easy to identify elements of the signal which depart from this permitted set. In such arrangements, reliance is placed on the known forms of the permitted signals.
SUMMARY OF THE INVENTION
The present invention seeks to provide a method of testing a line whilst in use for carrying live speech. A testing system is known in which the signal-to-noise ratio, or other measurable characteristics, of the system are determined by classifying samples as speech or as noise and comparing the properties of each sample. This is disclosed in a paper by David B Ramsden in IEEE "Globecom 91" pages 1761 to 1764, and in European patent 0565424. However, this does not attempt to measure the properties of the speech content itself.
According to a first aspect of the invention, there is provided a method of analysis of characteristics of a telecommunications system by measuring properties of a speech signal carried by a line under test, the method comprising the steps of: identifying a characteristic of the speech content of the signal received at a testing point which is naturally substantially invariant between individual talkers, and detecting deviations from that characteristic in the signal, thereby identifying properties imposed by the system on the signal.
According to a second aspect of the invention, there is provided a method of analysis of characteristics of a telecommunications system by measuring properties of a speech signal carried by a line under test, the method comprising the steps of: identifying a part of the speech signal having a property which varies in a predetermined relationship to a property of the original speech signal, detecting variations from that relationship in the received signal, and estimating the properties of the original speech signal therefrom.
According to a third aspect of the invention, there is provided a method of analysis of characteristics of a telecommunications system by measuring properties of a call carried by a line under t
REFERENCES:
patent: 5313555 (1994-05-01), Kamiya
patent: 5590242 (1996-12-01), Juang et al.
patent: 5664059 (1997-09-01), Zhao
patent: 5778336 (1998-07-01), Chou et al.
patent: 5794192 (1998-08-01), Zhao
patent: 5812972 (1998-09-01), Juang et al.
Yunxin Zhao, "A New Speaker Adaptation Technique Using Very Short Calibration Speech", Proc. ICASSP 93, vol. II, pp. 562-565, Apr. 1993.
Mazin G. Rahim et al., "Signal Bias Removal for Robust Telephone Based Speech Recognition in Adverse Environments," Proc. ICASSP 94, vol. I, pp. 445-448, Apr. 1994.
Yunxin Zhao, "An Acoustic-Phonetic-Based Speaker Adaptation Technique for Improving Speaker-Independent Continuous Speech Recognition," IEEE Trans. on Speech and Audio Processing, vol. 2, No. 2, pp. 380-394, Jul. 1994.
David B. Ramsden, "In-Service Nonintrusive Measurement on Speech Signals," Proc. GLOBECOM 91, pp. 1761-1764, Dec. 1991.
British Telecommunications public limited company
Hudspeth David R.
Smits Talivaldis Ivars
LandOfFree
Nonintrusive testing of telecommunication speech by determining does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Nonintrusive testing of telecommunication speech by determining , we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Nonintrusive testing of telecommunication speech by determining will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-325544