EP0725382A2 - Method and device providing digitally coded traffic information by synthetically generated speech - Google Patents

Abstract

The radio data system involves transmission of traffic information as synthetic speech. Various types of message sentence structures are available (1, 2, 3) and words from a dictionary (4) are combined (5, 6, 7). The results are processed (9, 10, 11) with the coded speech signals being subjected to a decoding operation (12) followed by speech generation (13) based upon control of a digital filter. The digital signals are converted (14) for input to an audio amplifier (15) coupled to a loudspeaker (16).

Description

The invention is based on a method according to the type of the main claim.

The radio data system (RDS) enables an additional and inaudible transmission of digital data in parallel to radio programs in a data channel. Specifications of the radio data system for VHF radio can be found in Tech. 3244 - E, March 1984 of the European Runfunk Union (EBU). Radio receivers with suitable RDS decoders can record and decode transmitted data in addition to radio reception with the same receiver. The data transmission takes place in 32 groups of 108 bits each, each group being assigned a specific service. Group 8a is intended for the transmission of digitally coded traffic reports.

The structure and coding of these traffic reports are specified in the standard proposal ALERT C, November 1990, published by the RDS ATT ALERT Consortium. The The essential elements of a traffic announcement are the location and the event. This information is cataloged, which means that a unique code is assigned to every traffic-relevant location and every traffic-relevant event. The chaining of the locations in the location table along existing streets shows the course. In addition to the usual facilities of a receiving device with an RDS decoder, facilities for decoding, storing, further processing and outputting the traffic reports are required to use the traffic message channel TMC (Traffic Message Channel).

For voice output, it is necessary to form sentences from the information received with the traffic reports and the information stored in the receiver that are immediately understandable to the user and sound largely natural.

The object of the present invention is to provide a method for outputting digitally coded traffic reports by means of synthetically generated speech, which enables this with the least possible amount of memory. Furthermore, with the method according to the invention, it should also be possible to take peculiarities of different languages into account.

This object is achieved in that, depending on the traffic message received, a carrier set is selected from a plurality of stored carrier sets and that location-based and event-related words and / or sentences are used in open positions of the carrier set, which are derived from the location codes and event codes. Through many different location codes and event codes as well as a large number of different language formulations for description The respective content of the traffic announcement results in a very large number of possible speech outputs spoken as sentences. With the method according to the invention, it is only necessary to store a sufficiently large number of carrier sets, the respective location codes and event codes, including the associated words.

An advantageous development of the method according to the invention is that the carrier sentences, the location-related words and the event-related words and / or sentences are stored in the form of phonetic transcription, that depending on the phonetic transcription from a phonetic catalog, sounds for the carrier sentences, for the location-dependent and for the event-dependent words and / or sentences are read out that the sounds forming a word are subjected to a transitional formation, that the carrier sentence with the words and / or sentences used is then subjected to a transitional formation between the words and digitally filtered and supplied with an acoustic reproduction becomes.

This further development has the advantage that an otherwise necessary implementation of the location-related and event-related words in phonetic transcription, which are shown, for example, in conventional script can be dispensed with.

The location-related words are usually stored in the language spoken in the respective area. However, if these words are inserted into the carrier sentences together with the event-related words and / or sentences, the emphasis in the language of the event-related words and the carrier sentences must be taken into account. Another development of the method according to the invention therefore provides that, together with the carrier sentences and with the event-related words and / or sentences, there is an associated one Accentuation history is saved, which is taken into account when reading the sounds from the phonetic catalog and that the location-related words are defined after the conversion into sounds.

This development takes into account that for logistical reasons it is usually advantageous to store location-related information in an exchangeable memory, for example a chip card, while the information dependent on the respective language, for example the carrier sentences and the event-related words and / or sentences, in another Memory are stored.

An adaptation of location-related words present in another language to the sound of the language of the carrier sentences is possible according to another development of the invention in that the location-related words are subjected to prosody conversion before insertion into the carrier sentences if they are in a different language when the carrier sets are filed.

With the Prosodie conversion it can be achieved, for example, that a motorist who is traveling in France and whose car radio outputs the traffic reports in German can hear the place name Paris (with the letter s at the end). However, names in other languages are also common for a number of locations, for example Aachen in the French language Aix la Chapelle. So that these special features can also be taken into account in the speech output, according to another further development it is provided that, in addition to the location-related words in the language of the respective area, the location codes also contain common location-related words in at least one other language, that in the language of the Carrier sets store location-related words inserted in the carrier sets and that the location codes for no location-related words are stored in the language of the medium sentences, the location-related words are used in the language of the respective area.

If the traffic announcements are also to be displayed visually, the method according to the invention can also provide that the location-related and event-related words and the carrier sentences are also stored as characters for display on an optical display device. In most applications, storage in the language of the respective area is sufficient, since these place names can be seen on traffic signs anyway, making comparison with the visual representation easier.

Despite ongoing advances in storage density, the storage capacity of removable storage will be limited or relatively expensive in the foreseeable future. In order to keep the required storage capacity of an exchangeable data carrier and possibly also an internal memory as low as possible, another development of the method according to the invention provides that frequently occurring components of location- and / or event-related words are stored as additional phonetic characters and that under the additional ones Phonetic characters in the phonetic catalog, the components are stored as allophones.

In an advantageous device for performing the method according to the invention for a radio receiver with a radio data signal decoder and a device for decoding the digitally coded traffic reports, it is provided that a speech synthesis device with memories for carrier sentences, event words or sentences and for location-related words is connected to the speech synthesis device the decoded traffic reports can be supplied and that an output of the speech synthesis device is connected to an input of an audio amplifier which controls a loudspeaker arrangement. With this device, a large number of different traffic reports can be output, the storage space required for this being relatively small.

An advantageous use of a location list with at least partially multilingual location-related words is possible with a device according to the invention in that a selection device is arranged between the memory for the location-related words and the speech synthesis device, which forwards the location-related words in that language to the speech synthesis device , in which the carrier sentences are removed from the memory, provided that the respective location-related words are stored in this language.

In a further development of the device according to the invention, foreign-language words can be adapted to the sound of the carrier sentences in that the location-related words fed to the speech synthesis device are subjected to prosody conversion if they are not available in the language in which the carrier sentences come from Storage can be removed. The prosody conversion can take place according to predefined rules or via catalogs.

Another development of the device according to the invention is that the carrier sentences and the event words or sentences are stored in several languages. As a result, a radio receiver with a multilingual voice output can be implemented.

The fact that the memory for the location-related words is arranged on an interchangeable data carrier creates a particularly favorable system in which device classes of different sound quality are possible which can use the same chip card. This is also particularly favorable because a distribution channel that is independent of the radio receiver can then be selected for the chip card.

Fig. 1

eine schematische Darstellung der Erzeugung von Sprachsignalen, welche vollständige Sätze umfassen, aus einzelnen empfangenen oder im Empfänger generierten Codes,

Fig. 2

ein Blockschaltbild eines Empfängers, der für die Ausgabe von Verkehrsmeldungen in einer Sprache ausgelegt ist,

Fig. 3

ein Blockschaltbild eines Empfängers, der für die Ausgabe von Verkehrsmeldungen einer Sprache ausgelegt ist, bei dem jedoch fremdsprachige Ortsbezeichnungen bezüglich der Sprachmelodie an die Sprache des Empfängers angepaßt werden und

Fig. 4

eine Blockschaltbild eines Empfängers, bei dem verschiedene Sprachen für die Ausgabe gewählt werden können.

Exemplary embodiments of the invention are shown in the drawing using several figures and are explained in more detail in the following description. It shows:

Fig. 1

1 shows a schematic representation of the generation of speech signals, which comprise complete sentences, from individual codes received or generated in the receiver,

Fig. 2

2 shows a block diagram of a receiver which is designed for outputting traffic reports in one language,

Fig. 3

a block diagram of a receiver, which is designed for the output of traffic reports of a language, but in which foreign language place names are adapted to the language of the receiver with respect to the speech melody and

Fig. 4

a block diagram of a receiver in which different languages can be selected for the output.

Identical parts are provided with the same reference symbols in the figures. The exemplary embodiments and parts thereof are shown as block diagrams. However, this does not mean that the circuit arrangement according to the invention is limited to implementation with the aid of individual circuits corresponding to the blocks. Rather, the circuit arrangement according to the invention can be implemented in a particularly advantageous manner with the aid of highly integrated circuits. In this case, digital signal processors can be used which, with suitable programming, carry out the processing steps shown in the block diagrams. The exemplary embodiments, together with further circuit arrangements within an integrated circuit, can form essential parts of a radio receiver.

A traffic report to be output using the method according to the invention has, for example, the form: A7, Kassel, in the direction of Hanover, between Derneburg and Hildesheim, traffic jam, 2 km.
Before this sentence is generated for the speech output, the following carrier set is selected from a multiplicity of stored carrier sets as a function of the various codes contained in the transmitted coded traffic announcement: A 〈1〉, 〈2a〉 direction 〈2b〉, between 〈2c〉 and 〈 2d〉, 〈3〉. The initially open positions are filled with placeholders, where word type 1 is a motorway number, word type 2 is a location description and word type 3 is an event description.

The event description is in turn carried out using a carrier set - also called an event set in the following - for example:
Traffic jam 〈4〉 km.
Word type 4 is a numerical value. For the sake of simplicity, all events are treated like carrier sets, even events without variable parameters, for example "Caution, a ghost driver is coming towards you". The carrier sets are stored and numbered in the radio receiver. According to the above-mentioned ALERT specification and the parameters one or more carrier sets selected, compiled and supplemented by the specific words instead of the placeholders.

In the exemplary embodiment shown schematically in FIG. 1, a location list 1, an event sentence list 2 and a carrier sentence list 3 are present, in which the carrier sentences and words to be spoken are stored in phonetic transcription. A suitable phonetic alphabet is known under the name SAMPA and is suitable for several European languages. A word can be written using SAMPA in its typical prosody (word melody), for example in German or French. No human speaker is required to encode the words in SAMPA, so the encoding is speaker-independent.

All segments required for the words to be spoken are stored in a segment library 4 - also called phonetic catalog - and can be called up as a function of the phonetic spelling read from lists 1, 2, 3, which means that at 5, 6, 7 the respective phonetic transcription is replaced by Phonemes, diphones and allophones are done. To produce the segment library, the segments are spoken, recorded, numbered and encoded in a known digital form, quantized, stored and provided with an address by a human speaker. Coding methods have become known under the names RELP, CELP and LPC.

The sound quality of the segments depends on the type of coding and quantization. A distinction must be made between parameterized and non-parameterized procedures. In the case of parameterized methods, modifications can be carried out to a limited extent at the time of playback, e.g. B. the tonal height, the Segment duration. At 6 and 7, an emphasis curve stored in lists 2 and 3 is also taken into account for each event or for each carrier set.

In the exemplary embodiment according to FIG. 1, it is assumed that the location list 1 is stored on an exchangeable data carrier which is valid for an area and which can be used regardless of the language in which the traffic reports are to be output. For this reason, no emphasis curve is stored in list 1. The emphasis on the place names is therefore set subsequently in a process step 8 depending on the respective language.

At 9, 10, 11, the coded speech signals representing the locations, the event sentences and the carrier sentences are subjected to a transition shaping within the words, whereupon after a combination of the components of the traffic report, a transition shaping between the words at 12 takes place. After decoding the sounds stored as parameter values at 13, the voice is generated at 13 ′, for example by suitable control of a digital filter that simulates the human pharynx. At 14, a digital / analog conversion is carried out, whereupon the signals from an audio amplifier 15 are fed to a loudspeaker 16.

Of the radio receivers, only parts and functions are shown in FIGS. 2, 3 and 4 which are necessary to explain the speech output according to the invention. The signals of group 8a are fed to a traffic message processing device 21 at 22 from an RDS decoder (not shown). According to the ALERT specification mentioned above, this data essentially contains information about the location, the type, the expected duration, the size and the Direction of an event. This information is decoded and selected in the device 21, which is preferably implemented in the form of a microcomputer or a digital signal processor. A selection takes place according to the selection criteria supplied via an input unit 23, so that, for example, only those traffic reports are reproduced which relate to the entered road section in a given direction.

In order to decode the traffic reports, data stored in part 25 of the location list 1 is required, for example the locations closest to the location mentioned in the traffic announcement in the course of a street and the affiliation to an area. This and other information are stored in a location list on a chip card 24 and, like other information described later, can be read with the aid of a reading device in the receiver. In the example shown it can be seen that the location no. 4711 belongs to an area in 2009 and that the neighboring locations are 4710 and 4712.

The traffic messages decoded in this way are fed to a device 26 for speech synthesis, the output signals of which are fed to the loudspeaker 16 via the digital / analog converter 14 and the audio amplifier 15. The device 26 can also advantageously be implemented with the aid of a digital signal processor, to which a program 27 is available. As in FIG. 1, the data required for speech synthesis are fed from a location list 1 which is stored on the chip card 24, an event record list 2, a carrier record list 3 and a segment library 4. The synthesis then takes place according to the method described in connection with FIG. 1 in the language in which the event set list 2, the carrier set list 3 and the segment library 4 are created are. The place names are then used in the language common in the respective area.

In the list of places 1 there are not only the verbal place names but also numbers, for example in connection with the connection point 15. Since it is naturally important to hear the number of a connection point in the language in which the rest of the output takes place, in the Place list 1 stores the number 15 as a number. A special part of the segment library 4 is used for the speech synthesis, in which the pronunciation of the numbers is stored in the respective language of the recipient. The list 1 also contains the place names in the usual font, so that an optical voice output can also be carried out with the aid of the chip card 24, and in phonetic transcription in one or more foreign languages. However, the foreign languages are not evaluated by the receiver according to FIG. 2.

The receiver according to FIG. 3 is expanded compared to the receiver according to FIG. 2 in such a way that the location designations to be used in the carrier sets are linguistically adapted to the language of the receiver. For this purpose, two different measures are provided in the exemplary embodiment according to FIG. 3. On the one hand, in addition to the place names in the language of the respective area (endonyms), place names from other languages (exonyms) are also stored on the chip card, insofar as these are common. In the case of Cologne, for example in the French language Cologne. In these cases, the receiver according to FIG. 3 can use a selection device 28 to select the place name available in the language of the receiver. In the case shown, for example from the languages English, Dutch, German and French. No place name is stored under the language of the recipient, which is the case with the majority of places in a country the case will be, the location name available in location list 1 is used for speech synthesis.

A further measure for improving the speech quality consists in the fact that the place names supplied in a language other than the language of the recipient are fed to a prosody conversion 29. Such a conversion is known per se and changes the speech melody in the direction of the language of the recipient.

After switching on the receiver, through special operation or when exchanging the location list (e.g. changing the chip card), the voice synthesis is configured. A German-speaking recipient, for example, contains a chip card with a list of locations for France. After switching on, the receiver recognizes that it is a chip card with a place list for France, which also contains German-speaking place names. The recipient then selects the conversion of the French prosody into German prosody on the prosody converter. Furthermore, each time the location list is accessed, a general check is carried out to determine whether a location has a German name or not. If there is no German-language name, the "source prosody" (here French) is converted into the "target prosody" (here German).

4, the event set list and the carrier set list are available in several languages, namely in German, English and French as event set lists 2de, 2en, 2fr and carrier set lists 3de, 3en and 3fr. In addition, the segment library 4 'is expanded to include special segments of the added languages, all languages being spoken by the same speaker. This means that the traffic reports can be issued in several languages. With the selection device 28 and the prosody conversion 29, the place names contained in the place list can be adapted to the language selected in each case.

Claims (12)

Method for outputting digitally coded traffic reports using synthetically generated speech, the traffic reports containing location codes and event codes, characterized in that, depending on the traffic report received in each case, one carrier set is selected from a plurality of stored carrier sets and that in open positions of the carrier set, location-related and event-related words are used, which are derived from the location codes and event codes. Method according to Claim 1, characterized in that the carrier sentences, the location-related words and the event-related words are stored in the form of phonetic transcription, that depending on the phonetic transcription, sounds are read from a phonetic catalog for the carrier sentences, for the location-dependent and for the event-dependent words are that the sounds forming a word are subjected to a transitional shaping, that the carrier sentence with the words used is then subjected to a transitional shaping between the words and digital filtering and is fed to an acoustic reproduction. A method according to claim 2, characterized in that together with the carrier sentences and the event-related words, an associated stress curve is stored, which is taken into account when reading the sounds from the phonetic catalog and that the location-related words are defined after the translation into sounds . Method according to one of claims 2 or 3, characterized in that the location-related words are subjected to prosody conversion before being inserted into the carrier sentences if they are stored in a language other than the carrier sentences. Method according to one of Claims 2 to 4, characterized in that, in addition to the location-related words in the language of the respective area, common location-related words are also stored in at least one other language under the location codes, so that the location-related words stored in the language of the carrier sentences are stored in the Carrier sentences are used and that the location codes for which no location-related words are stored in the language of the carrier sentences, the location-related words are used in the language of the respective area. Method according to one of Claims 2 to 5, characterized in that the location-related and event-related words and the carrier sentences are furthermore stored as characters for display on an optical reproduction device. Method according to one of the preceding claims, characterized in that frequently occurring components of location and / or event-related words are stored as additional phonetic characters and that under the additional phonetic characters in the phonetic catalog the Components are stored as allophones. Device for carrying out the method according to one of the preceding claims for a radio receiver with a radio data signal decoder and a device for decoding the digitally coded traffic reports, characterized in that a speech synthesis device (26) with memories (3, 2, 1) for carrier sentences, for event words or sentences and for location-related words, the speech synthesis device (26) can be supplied with the decoded traffic reports and that an output of the speech synthesis device (26) is connected to an input of an audio amplifier (15) which controls a loudspeaker arrangement (16). Device according to Claim 8, characterized in that a selection device (28) is arranged between the memory (1) for the location-specific words and the speech synthesis device (26), which selects the location-related words in that language to the speech synthesis device (26) forwards in which the carrier sentences are removed from the memory (3), provided that the respective location-related words are stored in this language. Device according to one of the preceding claims 8 or 9, characterized in that the location-related words supplied to the speech synthesis device (26) are subjected to a prosody conversion (29) if they are not in the language in which the carrier sentences are stored (3) can be removed. Device according to claim 9, characterized in that the carrier sentences and the event words or sentences are stored in several languages. Device according to one of claims 8 to 11, characterized in that the memory (1) for the location-related words is arranged on an exchangeable data carrier (24).

Images