Data processing: speech signal processing – linguistics – language – Speech signal processing – For storage or transmission
Patent
1997-08-11
2000-11-07
Hudspeth, David R.
Data processing: speech signal processing, linguistics, language
Speech signal processing
For storage or transmission
704228, 704225, G10L 2102
Patent
active
061449369
DESCRIPTION:
BRIEF SUMMARY
This application is the national phase of international application PCT/FI95/00667, filed Dec. 5, 1995 which was designated the U.S.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to decoding a received signal in a receiver of a digital communication system. The invention particularly relates to handling of speech coding parameters obtained from a channel decoder prior to the actual decoding to be carried out in the speech decoder.
2. Description of Related Art
In systems transmitting speech in digital form, such as mobile radio systems, the speech signal is first coded by some method prior to channel coding and transmission on the radio path. Speech that is digitalized in the speech coding is handled on a frame-by-frame basis in time intervals of approximately 20 ms by utilizing various kinds of methods, resulting in a frame specific set of parameters which characterize the speech. This information, i.e. the set of parameters, is channel coded and transmitted to the transmission path. In the channel coding, the information is protected by various kinds of error correction codes.
Upon receiving a radio signal, the signal is first channel decoded, whereby the original speech frames are obtained unless the transmission channel has caused disturbances. Speech frames of good quality may directly be applied to the speech decoder and used for speech synthesization. As is well known, the radio path is more or less under disturbance, which means, also as far as communication systems transmitting speech over the radio path are concerned, that some of the transmitted speech frames may as a result of the transmission errors prove to be erroneous at the reception to such an extent that they cannot be used for speech synthesization. Even if such codes that efficiently correct errors were used to protect the information to be transmitted, the received parameter values may still include serious transmission errors. The most significant parameters to be transmitted are, for the above reason, protected not only by the aforementioned error correction code but also by an error protection code, such as a CRC check sum (Cyclic Redundancy Check), capable of detecting errors. The check sum coding is capable of detecting speech frames that are badly corrupted in the transmission channel, and, consequently, it is possible to block their use for speech synthesization. The blocking procedure is imperative in order for the badly corrupted speech parameters not to cause heavy interference, e.g. clicks, in the synthesized speech. Upon detecting that a speech frame has been received so badly corrupted in the transmission channel that it is no longer possible to use the frame for speech synthesization, the received frame is classified as "a bad frame". The bad frames are separated from the rest of the speech frames, which are consequently called "good frames". Under extreme conditions on the radio path, situations emerge during which several successive speech frames received must be rejected and speech frames might even be lost over a period of several seconds.
In present-day digital cellular communication systems, such as the GSM (Groupe Special Mobile), an efficient error correction coding is utilized, enabling transmission of speech with good quality even over a bad quality connection. When the error correction capability of the error correction finally at some stage is exceeded, the received speech frames can no longer be used for speech synthesization. In such a case, a substitution method for bad frames is initiated, utilizing the aforementioned classification of speech frames as good and bad ones.
In the substitution process, the parameter values of bad frames are substituted for parameter values that cause as little audible disturbance as possible. A method proved to be proficient is to substitute all the parameter values of a bad, frame for the parameter values of the last received frame that is classified as good, and to feed the speech coder repetitively with the substitution frame. By r
REFERENCES:
patent: 4688224 (1987-08-01), Dal Degan et al.
patent: 5142582 (1992-08-01), Asakawa et al.
patent: 5255343 (1993-10-01), Su
patent: 5293449 (1994-03-01), Tzeng
patent: 5309443 (1994-05-01), Schorman
patent: 5333153 (1994-07-01), Brown et al.
patent: 5537509 (1996-07-01), Swaminathan et al.
patent: 5557639 (1996-09-01), Heikkila et al.
Jarvinen Kari
Vainio Janne
Azad Abul K.
Hudspeth David R.
Nokia Telecommunications Oy
LandOfFree
Method for substituting bad speech frames in a digital communica does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for substituting bad speech frames in a digital communica, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for substituting bad speech frames in a digital communica will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1651441