Nec Corporation

Nippon Electric Co., Ltd.

Shinichi Aikoh

Takao Nishitani

Akira Kanemasa

Atsushi Tomozawa

Kunihiko Niwa

Minoru Shinta

Input signals are processed to generate a plurality of signals having different features according to whether the input signals are speech signals or voiceband data signals, and these plural signals are entered into a neural network to be determined whether they have features closer to those of speech signals or of voiceband data signals. The classifying function of the neural network is achieved by inputting samples of speech signals and voiceband data signals and learning how to obtain correct classification results.

Speech/voiceband data discriminator

In a time division multiplexing method or device wherein first through N-th-channel samples are extracted at first through M-th unit intervals which correspond to first through M-th bit rates of first through N-th-channel input signals, where M is equal to or less than N, a control signal is produced in response to the input signals to indicate first through M-th consecutive fields in each of successive frames of a time division multiplexed signal and to indicate whether the samples are significant or insignificant samples in each frame. The first through the M-th fields correspond to the unit intervals and have a predetermined total time duration in each frame. The control signal is used moreover to arrange the samples in the fields of each frame by selecting the significant samples with preference to the insignificant samples. Together with the control signal, a sample arrangement information signal is produced to indicate arrangement of the samples in each frame and is arranged in an information field which remains in each frame. Depending on the number of the samples arranged in each field with the unit interval corresponding to the field under consideration, time durations of the fields may vary from frame to frame.

Time division multiplexing method with channel compression and a device

In a digital time division multiplexed telephone transmission system, a plurality of telephone signals are cyclically sampled and quantized so that each sampled value may be encoded into a binary codeword of a preset number of bits and that a series of codewords equal in number to the telephone signals may constitute a frame and a series of a predetermined number frames may in turn constitute a superframe. At the transmission end, an input circuit (202) receives the time division multiplexed telephone signals, and a code converter (204) coupled to the input circuit and responsive to a state-representing signal successively converts each of the binary codewords into a plurality of modified codewords. The modified codewords are variably assigned comparatively smaller and greater numbers of binary digits depending on the state-representing signal, the comparatively greater number being equal to or smaller than the preset number. A selection circuit (207) is supplied with the modified codewords and selectively allows one of the modified codewords to pass. A bit-number-assigning circuit (205) is coupled to the code converter and responsive to the output of the selection circuit to produce the state-representing signal in response to the corresponding one of at least one preceding frame, and for each of the modified codewords, the bit-number-assigning circuit produces a bit-number-representing signal indicating the number of bits assigned to the binary codeword being transmitted. A multiplexer (206) is coupled to the bit-number-assigning circuit and arranges in time domain the output of the selection circuit, bit-number-representing signal and a superframe synchronization signal. At the reception end, a code-inverse converter (1003) is responsive to the bit-number-representing signal to reproduce the binary codewords from the modified codewords.

Growing community of inventors

Tokyo, Japan

Shinichi Aikoh