Abstract: A method and apparatus are described that allow inexpensive speech recognition in applications where this capability is not otherwise feasible because of cost or technical reasons, or because of inconvenience to the user. A relatively simple speaker independent recognition algorithm, capable of recognizing a limited number of utterances at any one time, is associated with the base unit of an electronics product. To function, the product requires information from an external medium and this medium also provides the data required to recognize several sets of utterances pertinent to other information provided by the external medium.

Abstract: A spoken word or phrase recognition device. The device does not require a digital signal processor, large RAM, or extensive analog circuitry. The input audio signal is digitized and passed recursively through a digital difference filter to produce a multiplicity of filtered output waveforms. These waveforms are processed in real time by a microprocessor to generate a pattern that is recognized by a neural network pattern classifier that operates in software in the microprocessor. By application of additional techniques, this device has been shown to recognize an unknown speaker saying a digit from zero through nine with an accuracy greater than 99%. Because of the recognition accuracy and cost-effective design, the device may be used in cost sensitive applications such as toys, electronic learning aids, and consumer electronic products.

Abstract: A spoken word or phrase recognition device. The device does not require a digital signal processor, large RAM, or extensive analog circuitry. The input audio signal is digitized and passed recursively through a digital difference filter to produce a multiplicity of filtered output waveforms. These waveforms are processed in real time by a microprocessor to generate a pattern that is recognized by a neural network pattern classifier that operates in software in the microprocessor. By application of additional techniques, this device has been shown to recognize an unknown speaker saying a digit from zero through nine with an accuracy greater than 99%. Because of the recognition accuracy and cost-effective design, the device may be used in cost sensitive applications such as toys, electronic learning aids, and consumer electronic products.

Abstract: A method and apparatus is disclosed which allows the sending or receiving of either speech or digital data calls over a phone line by correctly connecting either a digital data machine or a voice phone with the line without any human involvement. An analyzer connected to the phone line interrogates each incoming call to determine if it is a voice call or a digital data call. If it is a voice call, the analyzer rings the phone and connects it to the line when the phone is answered. If the incoming call is a digital data call, the data machine, e.g., fax, is connected to the phone line. The distinction between voice and digital data calls is based in part on analyses of incoming response to an interrogation of the caller with messages from a speech synthesizer. For outgoing calls, the analyzer determines which of the phone and the data machines becomes active, and connects the active one with the phone line while it blocks access to the line by the other one until the outgoing call is complete.

Abstract: A storage device for use with a synthesizer of original information bearing time domain signals from compressed information time domain signals produced by predetermined different signal compression techniques. The storage device contains compressed information time domain signals and instruction signals specifying the particular compression technique applied to the original information bearing time domain signals to produce corresponding portions of the compressed information time domain signals. The compressed information time domain signals comprise a plurality of samples resulting from the predetermined signal compression techniques, the number of the different signal compression techniques applied to the original signals being greater than two and the ratio of the plurality of the samples to the minimum number of samples required to uniquely and intelligibly identify the original information bearing signals being no greater than about 0.2.

Abstract: Compression and synthesis techniques and related apparatus for time domain signals, particularly signals whose information content resides in the power spectrum such as speech and more particularly signals whose amplitude is aperiodic, such as unvoiced speech sounds. Compression techniques include eliminating serially redundant segments of information. Synthesis, particularly of unvoiced sounds which are sensitive to injected artificial periodicity, involves repeating sequentially portions of the same segmentrepresentative of the source signal, including commencing and terminating at different points each repetition, varying the length of the portion and reproducing the portion forwards and backwards in time. The invention finds application in speech compression and compact speech synthesis devices.

Abstract: Compression and synthesis techniques and related apparatus for time domain signals, particularly signals whose information content resides in the power spectrum such as speech. Compression techniques include adjusting the phase of harmonic components of a signal unit to obtain an equivalent power spectrum signal of a minimum number of discrete levels. The invention finds application in speech compression and compact speech synthesis devices.

Abstract: A method and apparatus for analyzing and synthesizing speech information in which a predetermined vocabulary is spoken into a microphone, the resulting electrical signals are differentiated with respect to time, digitized, and the digitized waveform is appropriately expanded or contracted by linear interpolation so that the pitch periods of all such waveforms have a uniform number of digitizations and the amplitudes are normalized with respect to a reference signal. These "standardized" speech information digital signals are then compressed in the computer by subjectively removing preselected relatively low power portions of the signals by a process termed "X period zeroing" and by discarding redundant speech information such as redundant pitch periods, portions of pitch periods, redundant phonemes and portions of phonemes, redundant amplitude information (delta modulation) and phase information (Fourier transformation).

Abstract: A speech synthesizer including a device for storing compressed digital signals corresponding to original information speech or audio waveform time domain signals, the digital signals including information signal portions and instruction signal portions identifying particular compression techniques applied to associated information signal portions; an output terminal for manifesting analog electrical synthesized signals corresponding to the original signals; a digital-to-analog converter having an output coupled to the output terminal and an input; and an intermediate signal processing circuit having an input coupled to the storage device and an output coupled to the digital-to-analog converter for expanding the information signal portions in accordance with the instruction signal portions to produce digital synthesized signals to be converted to analog synthesized signals by the digital-to-analog converter.

Abstract: This delta modulation system uses the differences between three successive samples of a waveform (the i-1, i, and i+1) such that the number of bits per digitization is reduced from four to two. The value vi of the ith sample is calculated by the equation v.sub.i =v.sub.i -1+f(.DELTA..sub.i-1, .DELTA..sub.i), and f is a special function defined by 16 values stored in a read-only memory.

Abstract: A method and apparatus for analyzing and synthesizing speech information in which a predetermined vocabulary is spoken into a microphone, the resulting electrical signals are differentiated with respect to time, digitized, and the digitized waveform is appropriately expanded or contracted by linear interpolation so that the pitch periods of all such waveforms have a uniform number of digitizations and the amplitudes are normalized with respect to a reference signal. These "standardized" speech information digital signals are then compressed in the computer by subjectively removing and discarding redundant speech information such as redundant pitch periods, portions of pitch periods, redundant phonemes and portions of phonemes, redundant amplitude information (delta modulation) and phase informaton (Fourier transformation). The compression techniques are selectively applied to certain of the speech information signals by listening to the reproduced, compressed information.