EP0057853B2 - Electronic system for delivering messages - Google Patents

Abstract

1. Electronic text transmitter for emitting short texts, e. g. announcements, in the form of analog electrical signals, in which text transmitter the short texts are adapted to be composed of text portions of different lengths of time which are stored, in digital form, in a set-value memory, said text transmitter comprising a control device to which the set-value memory is connected through a memory address counter, and further comprising a digital/analog converter following said set-value memory, characterized in that, to the output of the set-value memory (3), an end-of-text-portion decoder (8) is connected for the recognition of a mark contained in the digital data and being associated with each end of a text portion, said decoder emitting an end-of-text-portion signal, that, for generating predetermined speech intervals between two text portions, a speech-interval decoder (9) is connected to the output of the memory address counter (2) for the recognition of begin-of-interval marks or end-of-interval marks, respectively, emitted by said memory address counter (2), whereby between a begin-of-interval mark and an end-of-interval mark, the emission of data from the set value memory (3) to the digital/analog converter (5) is interrupted until the speech-interval decoder (9) emits an end-of-interval signal, and that the outputs of the end-of-text-portion decoder (8) and of the speech-interval decoder (9) are connected, through an OR-gate (0), with the control device (1) for the further performance of the run-down of the programme.

Description

The invention relates to an electronic text generator for the delivery of short texts.

Such text providers are generally known. They have been used, for example, for voice output over the telephone network, for example for order processing in retail or for answering machines. They can be controlled from a data processing system, for example a microcomputer.

The text to be stored in the read-only memory can be converted into digital voice data in accordance with the known method of delta modulation. The digital-to-analog converter downstream of the read-only memory then contains a delta modulator.

A fundamental problem of the text generators of this type is that they still require a relatively large amount of storage space in the read-only memory.

This makes the facilities technically and costly complex, which is particularly noticeable when it comes to smaller facilities for the delivery of one or more short texts, such as those that can occur with an answering machine, for example.

A device is known (IBM Technical Disclosure Bulletin Volume 19, No. 6, Nov. 1976, p. 2357, 2358) in which speech is recorded via an encoder and is stored digitally in a speech memory. This language can then be retrieved from the memory using a decoder and output via an analog output. Natural speech pauses that occur during the recording are counted by means of a separate pause counter, and a value characterizing the pause length is stored in the memory each time a speech pause occurs. When the texts are reproduced by the decoder, a separate pause down counter is used based on the value stored in the memory, which characterizes the pause length. This known method can only be used if only the speech pauses that occur when recording a continuous text are to be recognized, recorded and retrieved.

In the case of the electronic text generator mentioned at the outset, the texts are to be composed of stored parts of the text which are not necessarily only the text sections resulting from natural language pauses. Rather, they are deliberately selected text fragments, with parts of the same text that appear several times in a text or in several texts being saved only once and then used at various points in the text to be submitted. With such a method, it is not possible to store a signal containing the pause length in the read-only memory without great effort, since very different pause lengths can be connected to a certain text part, depending on its arrangement in the text.

The object on which the invention is based is to design a text generator of the type mentioned at the outset in such a way that short texts of various types and compositions can be given with a minimum of memory spaces.

This task is solved by an electronic text generator for the delivery of short texts, for example announcements, in the form of analog electrical signals, in which the short texts can be composed of text parts of different lengths of time stored in digital form in a read-only memory in such a way that identical text parts that are repeated several times a text or several texts, are only saved once and are then used at various points in the text to be delivered, with a control device (1) to which the read-only memory (3) is connected via a memory address counter (2), one the read-only memory ( 3) downstream digital-to-analog converter, and a speech pause decoding device (9) for recognizing pause start markings and in which a counter is used to generate predefined speech pauses, during its execution the data output of the read-only memory during a predetermined number of counting steps the D Digital-to-analog converter is interrupted, a text part end decoding device (8) for recognizing inaudible markings in the digital data which are assigned to each text part end and are stored in the digital data and which emitting a text part end signal being connected being connected to the output of the read-only memory , the speech pause decoding device (9) is connected to the output of the memory address counter (2) and recognizes the pause start and pause end markings, which are output by the memory address counter under program control, the generation of the speech pauses between a pause start and pause end mark Memory address counter (2) operates during the pause in a predetermined address range in which the read-only memory (3) is not equipped and the speech pause decoding device (9) emits a pause-end signal and the outputs of the text-part decoding device (8) and the speech pause Decoding device (9) via e are connected in OR gate (0) to the control device (1) for advancing the program sequence.

The ends of all parts of the text that compose the entire text are therefore identified by inaudible markings in the digital data. These markings are expediently the same for all text parts and are not bound to a specific data storage address. This enables an individual definition of the text part lengths that corresponds exactly to the respective text part. The markings are recognized by the end of text decoding device, which a corresponding end of text Si generated. The language breaks are treated like parts of text when the program is created. However, they are not stored in the read-only memory, but an address range is specified for the duration of the pause in which the read-only memory is not populated. The speech pauses can be generated by presetting certain pause start addresses in the memory address counter, which are recognized by the speech pause decoding device which emits a pause and pause signal. The end of text signal and the end of pause signal are used to further control the program flow.

Since no data is supplied to the digital-to-analog converter from the read-only memory during the pauses in speech, it is expedient if, according to claim 2, the speech-pause decoding device emits a control signal during a predetermined speech pause, by means of which the input of the digital-to-analog converter has a Device for delivering a fixed input signal is connected so that it receives the digital input data during the pauses in speech, which generate the NF level "0" at the converter output.

A further particularly advantageous embodiment of the text generator according to the invention is the subject of claim 5. It is expedient to monitor the low-frequency signal emitted by the digital-to-analog converter in order to immediately detect any faults which, for example, can lead to text mutilation. By coupling this monitoring of the output low-frequency signal with the speech pause decoding device, a particularly sensitive monitoring of the low frequency is possible, which, since it does not respond to the specified speech pauses, can be designed so sensitive that it detects even the slightest interruptions in the low frequency Signal in the range of approx. 60 milliseconds reacts.

An exemplary embodiment of the text generator according to the invention is explained in more detail below with reference to the drawing.

In the drawing, only the parts of the text generator that are essential to the invention are shown in a block diagram.

The program sequence is controlled by a control device 1 which, depending on the purpose of the text generator, can be connected to other monitoring and control devices, for example a microprocessor. All of the texts to be submitted, e.g. Announcements can be made, occurring text parts are digitized according to the known principle of delta modulation and stored in the read-only memory 3 (e.g. ROM, EPROM) one after the other. The start addresses of these text parts stored in the read-only memory 3 are defined in a known manner in the control program and are given by the control device 1 to the memory address counter 2. The ends of all parts of the text are identified by inaudible marks in the digital data. These marks are the same for all text parts and are not tied to a specific data storage address. The text parts are output in such a way that the memory address counter 2 is preset to the fixed value memory start address of any text part and switches from this counter reading, controlled by a clock signal from the clock generator 12 with the frequency f6. In this way, the content of one memory cell after the other is released and fed to the digital-to-analog converter 5. The read-only memory 3 is constructed in a known manner in a byte-wise organization, i.e. 8 bits are given simultaneously to 8 outputs under each address. For this reason, a parallel-to-serial converter 4 must be switched on before the digital-to-analog converter 5, which can only process 1-bit serial information.

The marking occurring at the end of a text part is recognized by the text part end decoding device 8 connected to the read-only memory 3 and a corresponding text part end signal is supplied to the control device 1 and the memory address counter 2 via an OR gate 0. A text section can either be followed directly by another text section or a language break.

Predefined speech pauses between individual text parts are generated in such a way that, according to the program entered, the memory address counter 2 is set to an address within a predefined address range in which the read-only memory 3 itself is not populated. The procedure can be such that a certain fixed value in the memory address counter 2 is assigned to the end of the pause, which is the same for all speech pauses. Pauses of different lengths can then be realized by programming different start addresses within this area. In this way, longer speech pauses can be achieved by lining up several pause sections. The pause start and pause markings emitted by memory address counter 2 are recognized by the speech pause decoding device 9 connected to the memory address counter 2. A speech pause end signal is generated, which is likewise supplied to the control device 1 and the memory address counter 2 via the OR gate 0.

In order to compose a certain text from the text parts stored in the read-only memory 3 and the speech pauses that may occur between them, the text part start addresses and speech pause start addresses are written into the program of the controller 1 in the desired order. At the start of the text and each time a part of the text is recognized by the text part-end decoding device 8 or a speech pause end by means of the speech pause decoding device 9, the memory address counter 2 is preset by the controller 1 to the next text part start address that follows in the program.

The digital-to-analog converter 5, which is constantly clocked by the clock generator 12 with a clock signal of the frequency f3, emits an analog output signal which runs through a playback filter 6 and is thereby limited to the frequency band relevant for voice announcements, which also applies to the Digitization inevitable quantization noise is dampened. The low-frequency signal reaches the output NF-A via a downstream amplifier 7.

So that at the input of the digital-to-analog converter 5 during the pauses in which no data is transmitted from the read-only memory 3, there are no undefined states which can lead to interference signals at the converter output, the pause decoder 9 emits a control signal during the pauses in speech, by means of which the input of the digital-to-analog converter 5 is connected to an output of the clock generator 12 during the speech pauses by means of a switch S, from which a clocked input signal of frequency f4 is supplied from the digital-to-analog converter 5. This input signal expediently has the frequency f4 = ½ f3.

A program monitoring device 10 controls the program sequence and transmits the control signals to the memory address counter 2 via an AND gate U, to which the clock signal f6 is supplied at the same time. Program ends are indicated via the P-Ü output.

A device 11 for monitoring the low-frequency signal fed to the output NF-A indicates a fault in the delivery of this signal via the output NF-Ü. The device 11 can be designed to be very sensitive to a failure of the low frequency signal monitoring device, because it is only released when "text" is programmed, i.e. during the predetermined speech pauses it is switched off by a control signal emitted by the speech pause decoding device 9. In this way, monitoring of the low-frequency signal which is emitted is carried out very quickly.

Claims (5)

1. Electronic text transmitter for emitting short texts, e.g. announcements, in the form of analog electrical signals, in which text transmitter the short texts are adapted to be composed of text portions of different lenghts of time which are stored, in digital form, in a set-value memory, such that identical text portions which occur more than once in a text or in several texts are stored once only and are then inserted at various points in the texts to be transmitted, said text transmitter comprising a control device (1) to which the set-value memory (3) is connected through a memory address counter (2), and further comprising a digital/analog converter following said set-value memory (3), and further comprising a speech-interval decoder (9); for the recognition of begin-of-interval marks, and in which text transmitter a counter is used for generating predetermined speech intervals, with, when said counter runs down, the emission of data from the set-value memory to the digital/analog converter being interrupted during a predetermined number of counting steps, whereby, to the output of the set-value memory, an end-of- text-portion decoder (8) is connected for the recognition of non-audible marks contained in the digital data and being associated with each end of a text portion and being stored in the set-value memory (3), said decoder emitting an end-of- text-portion signal, with the speech-interval decoder (9) being connected to the output of the memory address counter (2) and detecting the begin-of-interval and end-of-interval marks emitted in program-controlled manner by the memory adress counter, whereby, for the generation of the speech intervals between begin-of-interval marks and end-of-interval marks, the memory address counter (2) works during the speech interval in a predetermined address area in which the set-value memory (3) is not equipped and the speech-interval decoder (9) emits an end-of-interval signal and the outputs of the end-of-text- portion decoder (8) and of the speech-interval decoder (9) are connected, through an OR-gate (0), with the control device (1) for the further performance of the run-down of the program.

2. Electronic text transmitter as claimed in claim 1, characterized in that the speech-interval decoder (9), during the occurance of a predetermined speech interval, emits a control signal by means of which the input of the digital/analog converter (5) is connected with a device (12) for emitting a fixed input signal.

3. Electronic text transmitter as claimed in claim 2, characterized in that the device for the emission of the fixed input signal is a clock generator (12).

4. Electronic text transmitter as claimed in claim 3, characterized in that the output frequency of the clock generator (12) which is transmit- tered to the digital/analog converter (5), is f4 = ¹/₂ - f3, whereby f3 is the clock frequency which is continuously transmitted to the digital/ analog converter (5).

5. Electronic text transmitter as claimed in any of claims 2 to 4, characterized by a divice (11) for controlling the low-frequency signal emitted by the digital/analog converter (5) which device is switched off by the control signal of the speech-interval decoder (9) during each of the predetermined speech intervals.

Images