Abstract: A speech analyzing stage (12) and a method for analyzing a speech signal is described. The speech analyzing stage (12 ) is part of an automatic speech recognition system (10) and is adapted for analyzing in the spectral domain a speech signal sampled at one of at least two different system sampling rates.

Abstract: The invention relates to a method and a device for avoiding interruptions in voice transmission in a cellular communication system. Voice data are thereby divided into segments, and the segments are associated with transmission quanta, so-called bursts. A number of segments is coded in a first transmission mode and is made available for transmission. Said first transmission mode is, for instance, for a transmission at full rate. Additional segments, which follow, are subsequently coded in a second transmission mode and are made available for transmission. The second transmission mode is, for instance, for transmission at half rate. Through the change from a first transmission mode to a second transmission mode, parts of the associated transmission quanta remain unused by applying the so-called interleaving. Said unused transmission quanta are used for performing additional functions, such as initiating a hand-over or FACCH signaling.

Abstract: A mobile terminal (100) which is controllable by spoken utterances like proper names or command words is described. The mobile terminal (100) comprises an interface (200) for receiving from a network server (300) acoustic models for automatic speech recognition and an automatic speech recognizer (110) for recognizing the spoken utterances based on the received acoustic models. The invention further relates to a network server (300) for mobile terminals (100) which are controllable by spoken utterances and to a method for obtaining acoustic models for a mobile terminal (100) controllable by spoken utterances.

Abstract: Low rate data signals, such as standard TTY signals and standard DTMF signals, can be effectively analyzed and encoded by speech encoders employed in conventional digital cellular system. This is accomplished by providing a low rate data encoder in the transmission path of the low rate data signal, just prior to the speech encoder used by the digital cellular system. In so doing, the standard, low rate data signal is transformed (i.e., modified) by the low rate data encoder into a signal that is more compatible with the speech encoder. Similarly, a low rate data decoder is provided at the receiving end of the transmission path to decode the modified, low rate data signal.

Abstract: Alternate speech and data communication over a single circuit is implemented within a communication network which include first nodes which include both a data source and a speech source. Digital information, comprising either speech parameters or data, is sent in discrete blocks between the first node and a second node connected to the network. An inband signaling bit pattern is included within selected blocks sent by a TX-alternator in the sending node to indicate whether all following digital information is to be interpreted by the receiving node as speech or data and a RX-alternator in the receiving node interprets the digital information in accordance with the last received inband signaling pattern. One exemplary implementation is in a GSM cellular radio network in tandem free operation.

Abstract: The invention provides method and apparatus for transporting tone signalling information through a GCME link in a GSM network supporting TFO and speech compression modes for the passage of signals through the link, respectively using TFO frames and TRAU-like In a data stream to the local entrance to the GCME link, a tone signal to be transported is detected. The detected tone signal is coded into bits of a TFO or TRAU-like frame for sending through the GCME link, including setting bits to mark the frame as a containing tone signalling information and setting bits to identify the detected tone signal, whereby the tone signalling information is transported in-band through the GCME link as a marked TFO or TRAU-like frame. At the remote exit from the GCME link, the marked frame is detected and the tone signal regenerated in dependence upon the bits of the marked frame.

Abstract: A source/channel encoding mode control method in a TDMA radio communication system determines the current type of source signal to be encoded and transmitted, restricts encoding to a class of source/channel encoding modes compatible with the determined type of source signal, determines a quality measure for previously transmitted signals that have been received and decoded, and selects based on the quality measure, the most suitable source/channel encoding mode for the determined class.

Abstract: A power control system in a TDMA radio communication system has traffic channels which are associated with a set of speech/channel encoding modes. Each mode has a different mix of speech encoder bit rate and data protection bit rate but the same total available gross bit rate. The transmitter includes a power control unit that replaces a mode allocated to a channel by another mode having either a higher or a lower data protection bit rate and either a lower or a higher speech encoder bit rate if the sound to be encoded requires either a lower or a higher speech encoding bit rate. The power control unit also controls a power adjustment unit that either reduces or increases the output power of the transmitter to a lower or a higher level such that an estimated decoded speech quality measure at the receiver is substantially constant.

Abstract: Complex coding of speech signals, as this is used, for example, in the GSM half rate speech coder, requires the use of a signal processor. On the other hand, a hands-free facility requires an arithmetic unit which adaptively calculates filter coefficients. This is only effected during speech pauses in which the speech coder is switched to a mode in which it generates codes for background noise with only little processing circuitry and expense. According to the invention, the signal processor for the coding device is therefore used for calculating the coefficients for the echo cancelling filter during speech pauses. This requires the use of not more than a single signal processor which can thus be used twice.

Abstract: A signal processor which includes a first data processing unit for processing data supplied thereto by a data bus system which operates in successive command cycles, and at least one further data processing unit operating in parallel with the first processing unit to process data supplied from the first processing unit during the command cycles. Since the further processing unit does not receive data from the bus system, that processor and still further processors can be added without requiring extension of the bus system. The parallel operation of a plurality of processing units achieves enhanced speed of computation, so that the signal processor is particularly adapted to computation of auto-correlation and cross-correlation functions and for FIR digital filtering.

Abstract: When speech data are transmitted via radio in the European mobile radio telephone system according to the GSM standard, always 20 ms speech data are transmitted as source-coded code words between the mobile stations and the fixed stations. The fixed stations are again connected to a central station via a connection having a smaller bandwidth, in which central station the speech decoders are arranged. Since transmission errors which cannot be corrected by the channel decoder in prior-art system are masked in the channel decoder in the central station, only a very simple error masking option is provided. According to the invention, the channel decoder which is arranged in the mobile and fixed stations close to the receiver, more signals are derived which feature the reliability of the received data and, in dependence on these additional data, extensive different error masking measures are carried out on the received code words. This does not only take place in the mobile station, but also in the fixed station.

Abstract: In so-called Code Excited Linear Prediction (CELP) coding methods for speech signal transmission, a codebook look-up method is used which is very processor-intensive. To conserve power, during speech pauses not only the transmitter but also the speech coder is turned off substantially completely. Consequently, when the speech signal resumes there is a transition interval before the filters of the speech coder become adjusted to full operation. For this reason, according to the invention, the filters are not turned off during speech pauses but are directly driven by codebook excitation vectors which correspond to the speech signal then being processed. As a result, there is a smoother and hardly perceptible transition between background noise and the speech signal when the latter resumes. An artificial background noise is produced in the receiver during speech pauses.

Abstract: Digital signal processing often requires the fast summing of a chain of products. Known signal processors often use two separate dam buses via which the values to be multiplied are supplied in parallel, it being assumed that these values originate from different sources, for example from different memories. Because a product of two binary numbers has double the number of positions, therefore, an adder having double the word width is also used. In order to reduce this substantial expenditure at the expense of only a slight reduction in speed, an adder is provided having only the single word width and to process the most-significant and least-significant bits of the product during two successive clock periods. The values to be multiplied can then be supplied successively.

Abstract: In a radio communication system, in particular a mobile communication system, variable-rate information flows originating from different sources (CV1, CS1; CV2, CS2) and relative to a same communication are transmitted on a same radio channel. Each mobile station (M1 . . . Mh) and the fixed part (B1 . . . Bn; RNC) of the system comprise a unit for the control of the variable-rate transmission, which dynamically allocates the available bits to the different streams by taking into account the needs of the sources (CV1, CS1; CV2, CS2), the conditions of the channel (6) and the system occupancy. (FIG. 1).

Abstract: During algebraic summing of code words from conference participants to form a partial sum-code word overflow of the values may occur. Such an overflow can be prevented when prior to the adding operations the numerical range is extended and an adjustment of weighting factors is effected in dependence on the measured speech intensity of the partial sum-code word. By multiplying by these (attenuation) factors the outgoing subscriber-code word is mapped again on the original numerical range in the attenuation circuit. To reduce a noise level originating, for example, from line interferences, quantization errors, etc. subscribers which are currently not speaking can be attenuated in a further attenuation control circuit of the conference arrangement.