DE3518737A1 - Speech storage arrangement for a telecommunications terminal, in particular a telephone terminal - Google Patents

Abstract

The speech storage arrangement according to the invention has different memory areas in which a different speech coding is in each case stored. When information is output, the memory area with the coding which generates the highest speech quality is first read out. The memory areas with a coding for an inferior speech quality but with a lower storage requirement per time unit are then activated along with a speech synthesis device.

Description

Telenorma, Telefonbau und Normalzeit GmbH, ⁰ ^ IO / ο / Mainzer Landstrasse 128 - 146, 6000 Frankfurt am Main

Speech memory arrangement for a telecommunication *, in particular telephone terminal

The invention relates to a voice storage arrangement for a telecommunication terminal, in particular a telephone terminal according to the preamble of claim 1.

It is known for automatic voice output in telecommunications, in particular telephone terminal equipment for generating artificial speech, so-called synthesizers, to use. An application example of this type is described in DE-OS 31 11 798. Here, the ν, οη is a Answering machine to be given voice information stored in predetermined sentences so that these - in part after a previously Choices made - can be output one after the other as synthetic speech. It is also provided that Such texts can be completed with numerical information (date, time) are. In addition, there is a magnetic tape memory for storing the spoken messages coming from the caller Information available.

With such a facility, if a caller is immediately confronted with synthetic speech, the impression could be arise that there is a faulty connection, with some subscribers being inclined to immediately return the handset hang up. In any case, it is difficult for an inexperienced user to adjust to an artificial voice announcement, especially if the answer is expected from someone whose voice is known.

The object of the invention is to present an arrangement for several purposes, wherein the user gradually becomes accustomed to the different quality of a synthetic voice announcement. This is going through Features achieved as specified in claim 1.

This is achieved in an advantageous manner that the user an automatic voice announcement gradually becomes accustomed to the poor voice quality of a voice synthesis device, in that the beginning of an announcement is always taken from a coding that offers high speech quality. Included If necessary, it can be achieved that the relevant speaker can be recognized by his voice. If the announcement now contains the information that the information will be provided immediately or gradually in the further course of the process If a transition is made to a synthetic language, the listener is prepared for it, which enables him to understand the artificial language is made much easier. Further developments of the invention emerge from the subclaims.

An embodiment of the invention is based on the following a drawing explained in more detail. The one shown in the drawing Speech memory SPSP is divided into several memory areas SPB1 to SPBn. In addition to the digital language information Identifiers are also stored that indicate the way in which the speech information is coded. With one of the Control device SE influenceable address generator ADG, the memory areas SPB can be controlled one after the other. Included the control device SE takes the respective identifier KZ

JO, so that the assigned to the respective coding type Converter device D / A, Dec or SSY is controlled.

The one at the output of the converter device being controlled appearing analog speech information reaches an electroacoustic transducer EAW via an amplifier V, if As in the example shown, it is a question of the stored speech being listened to directly. The exit _ of the amplifier V can, however, also be connected to a connecting line if the speech information is through a remote caller should be answered.

ORIGINAL INSPECTED

oo Io / ο /

In relation to the same storage capacities within the storage areas SPB, there can be either a few Voice information with a high voice quality or a lot of voice information with a poor voice quality save - According to the application and the requirements however, the individual memory areas SPB can have different sizes. So it's a very big one Freedom of movement given, in terms of type and detail a language information.

Some application examples are described below, the mode of operation of the voice memory arrangement being described will. It was initially assumed that the voice storage arrangement is used in an answering machine.

The start of the announcement is stored in the first memory area SPB1, as is the case with the other applications which is therefore also activated first by the address generator ADG. The language information is there in a coding stored, with which a high speech quality is achieved.

For this purpose, for example, the so-called adaptive differential Pulse code modulation suitable, in which case the individual speech sample words are designed for this type of coding Digital / analog converter D / A are fed. Such an announcement could be, for example:

"Here is Mr. M's automatic answering machine, good day. You will now receive some information in artificial language. "

This announcement can of course also be handled in a more polite form if more storage capacity is available stands. With the coding type selected in the example, adaptive differential pulse code modulation ADPCM is the speech quality high enough to make the called party's voice recognizable.

When this announcement has ended, the next memory area SPB2 is activated, in which speech information, for example are stored according to the linear predictive coding LPC. The language information stored in this memory area SPB2

can, for example, provide information in this application case contain the reason for the participant's absence and the time when he can be reached again. aside from that Short instructions for the options available on the answering machine can be announced. So can the participant be advised that he can leave a message if this is provided in the answering machine.

In the other memory areas, coding can be used in cooperation with a speech synthesis device SSY permits further detailed information and explanations, although a poorer voice quality is output is, but only a small storage capacity per unit of time is required. The voice information encoded in this way may contain detailed explanations of the operation of the answering machine, if the user or the caller wants to listen to stored texts. These texts can be a control of the message text. It is but also possible that messages recorded by a caller can be specifically listened to if the Speech synthesis device SSY detailed operating instructions are followed.

Another application for the speech memory arrangement according to the invention results in sic-h if an information system, e.g. stock market news, provides detailed information on Should be available. As previously described, the announcement is initially provided with a short notice in high voice quality initiated, the content of which is stored in the first memory area SPB1. , This notice could read as follows: "You hear the latest stock market news in artificial language."

This is followed by the individual details with company name and share price of a lower language quality. Depending on the embodiment the information system can now go to each individual information an even more detailed explanation, for example about the trend and the previous status, can be given, with it is advisable to use a further reduced speech quality in order to save storage space. If at Not listening to the information and the detailed explanations

are desired »so there is the possibility of a To achieve a dialogue with the information system that, instead of the detailed explanations, it switches to next information. You can switch to a memory area, e.g. SPB2, with a higher voice quality will. In this case it is up to the user of the information system whether he quickly needs a lot of information less wants to get detailed or targeted detailed information.

In the same way, an acoustic operating manual or operator guidance for a device can be set up with the voice memory arrangement according to the invention. Here, too, the user is initially informed with a high level of speech quality which features are available, and that an explanation is now made in artificial language. The further, more detailed information about the features and their mode of action are stored in a further voice memory SPB with a lower voice quality. With a yes-no input, the user can decide whether or not he wants a very detailed explanation of certain features.

P 3964

Gö / we

5/20/85

- blank page -

Claims (1)

Telenorma, Te-l-efönbäu-'un-d "Norraaizeit GmbH, Mainzer Landstrasse 128 - 146, 6000 Frankfurt am Main 3618737 Speech memory arrangement for a telecommunication terminal, in particular a telephone terminal Claims 1. Speech memory arrangement for a telecommunication, in particular Telephone terminal, where the digital memory content is used for linguistic information output and the language is also generated synthetically, characterized in that, that the speech memory (SPSP) has different memory areas (SPB) in which one ^ O different language coding is stored, and that at the information output first of all the memory area (SPB1) with the highest speech quality generating Coding, e.g. ADPCM, is read out and then the memory areas (SPB2 to SPBn) with a coding for a lower speech quality but with a lower memory requirement per unit of time, wherein a speech synthesis device is activated. 2. Speech memory arrangement according to claim 1, characterized in that that difficult words or word sequences within a text that are in memory areas (SPB2 to SPBn) for the Speech synthesis is arranged to be stored with a coding for higher speech quality, and that then the language is generated by direct analog conversion. ORDINAL INSPECTED 3. Speech memory arrangement according to claim 1, J J Ib / ο / characterized, that several coding methods are provided for a step-by-step adaptation to the synthetic speech. 5 4. Speech memory arrangement according to claim 1, characterized in that that through a dialogue with the person listening in on the information can be effected that on memory areas (SPB) with a different coding is switched if the synthetic speech is not clearly understood , whereby short announcements with high speech quality are then given. 5. voice memory arrangement according to claim 1, characterized in that that in the speech memory (SPSP) identifiers (KZ) are stored, which indicate the coding method according to which the Voice output has to take place. 20th P 3964 Gö / we 5/20/85