EP2592623B1 - Technique for outputting an acoustic signal by means of a navigation system - Google Patents

Description

Technical area

The present disclosure relates generally to navigation systems. Concretely, a technique for outputting an acoustic signal by means of a navigation system is given as part of route guidance.

background

Navigation systems with voice output, especially for vehicles, have conquered the market. No modern navigation system is available without voice output. The advantages of speech output are obvious: any purely visual display, if not in the environment of a state-of-the-art head-up display (HUD) for vehicles, can distract the driver from the traffic. This is because the driver has to direct the view of the person lying in front of him slightly to the side or down (or, especially in portable navigation systems, upward) in order to record the information presented to him purely visually. It should be noted that even HUD-based systems typically have a voice output to minimize the distraction of the driver. Even lower distractions, such as the use of mobile phones without hands-free system, were banned by the German legislature.

As advantageous and desirable as voice output in navigation systems is, the technical implementation of text-to-speech (TTS) mechanisms is particularly difficult, especially in navigation systems with often system-limited processing resources. Almost every living language, no matter which language strain it belongs to, is a highly complex, not always stringently logical set of rules that can only be approximated by means of computer logic.

One possible prior art TTS approach is to use graphemes, ie, the typically smallest meaningful graphical units of the particular language's writing system. For example, a written word (eg, in the form of a text string) expressed by several graphemes can be converted into a spoken acoustic signal by one of the following mechanisms.

As used in this specification, the terms grapheme, phoneme, and allophone are each attributed the meaning content described below.

An allophone represents a variant of a phoneme in the sense of this description, i. Allophones are different implementations of a phoneme. In contrast to phonemes, the allophones have the ability to reproduce nuances for different speakers, and then pronounce the same phoneme differently (speaker-dependent) as an allophone. An example is the rolled and the rubbed [r]. A high German speaker would pronounce the phoneme / r / as a rubbed 'r', whereas a dialect-colored speaker would realize the phoneme / r / as a rolling 'r', each with the same given phoneme sequence.

In summary, the phoneme is an abstraction of a concrete sound statement. To be more precise, it is a "class of sounds ... that share all distinctive characteristics, in which non-distinctive ones can differ" (see Meibauer, Jörg, Introduction to German Linguistics, 2nd ed. , P. 83).

As described, concrete implementations of phonemes may differ significantly from each other and yet be assigned to the same phoneme. The realizations (instances) of a phoneme are also called allophones (see entry "phoneme" in Wikipedia ™).

The phoneme is to be distinguished from the grapheme. The grapheme is the smallest meaningful unit of written language. The phoneme, on the other hand, is the smallest meaning-discriminating unit of the spoken language (see entry "Phonem" in Wikipedia ™).

In contrast to the phonemes, a morpheme has both a sound and a graphic expression. In other words, the phoneme is different from the morpheme. The phoneme is defined as the smallest semantic meaning, the morpheme is defined as the smallest meaningful linguistic entity. A morpheme usually consists of one or more phonemes (or in the written language Graphemen) (see entry "phoneme" in Wikipedia ™).

As an example, the morpheme called 'forest'. The plural is 'forests' and yet comes from the same morpheme as forest. This is exactly where the morpheme is different from the phoneme, which is focused exclusively on pronunciation. That is, the morpheme has a semantic (meaningful) component.

Fig. 1 FIG. 12 shows a graph-based speech output system 100 that includes a database 1001, a speech synthesizer 1003, and an acoustic output device 1002. The speech synthesis device 1003 includes the functions grapheme to phoneme converter (g2p) and acoustic signal generation (wave-gene). The database 1001 can store graphemes in various languages, represented are German, Polish and Czech.

First, in step g2p, a conversion into phonemes (smallest meaning-discriminating unit of the audible speech) ensues, then in step wave-gen the speech synthesizing. In step g2p, both automatic methods that determine pronunciation, stress, and duration, as well as manual methods that use manually created exception dictionaries for sentences, words, and word parts may be used. In step wave-gen, an acoustic signal is generated from a phoneme.

The problem here is that there are different phoneme formats, for example, depending on the manufacturer of the speech synthesis software. Some phoneme formats are language-dependent, ie only usable if the converter is set in step wave-gen to the language suitable for the phoneme.

Fig. 2 1 shows a phoneme-based speech output system 100 that includes the database 1001 and the wave-gen function described above (here as the sole component of the speech synthesis device 1003) and the acoustic output device 1002. As in Fig. 2 shown, database 1001 contains not only the written place and street names (as graphemes), but also the corresponding phoneme sequences, both in different languages, representing German, Polish and Czech. These phonemes or phoneme sequences generally have better quality than those which an automatic converter (see step g2p above) can generate. Both graphemes and phonemes or phoneme sequences are regularly stored in the nominative.

For a textual representation on the screen of a navigation system, the nominative is usually sufficient. For using place and street names in one spoken phrase arise here but problems: "Please take the second exit of the main square" is grammatically wrong because the nominative was used instead of the genitive.

A common solution is to circumvent the problem, namely to simply change the sentence: "Please take the second exit at the main square" solves the problem in German, since the dative required here (coincidentally) corresponds to the nominative stored in the database , The disadvantage is that on the one hand not always the conversion leads to the desired result or no solution is possible and thus not all case grammatically correct pronounced. In such situations, the approximation to the proper pronunciation is assessed and the lower quality accepted, or another method (such as the methods described above) chosen.

Fig. 3 shows a multi-conversion voice output system 100. The database 1001 and the acoustic output device 1002 are equipped with the corresponding elements Fig. 2 identical. The speech synthesis device 1003 includes other functions described below.

The multiple conversion represents an alternative solution in which a set of rules is created on the basis of graphemes, for example: genitive rule: *** becomes *** places. This control system is particularly suitable for Eastern European languages, as there are more cases that prohibit the nominative in route guidance announcements.

• \ (.*/)' Plak \(.*/) --> \1Pla'Ku/2
• \ (.*/)Na'mezti \(.*/) --> \lNa'meztim/2

The desired graphemes (eg main square in the nominative) are in the step reg-expr. declined by means of a set of rules in the appropriate case (eg genitive). However, this set of rules can only be created by experts with appropriate language and procedural knowledge and is therefore expensive, eg:

• \ (. * /) 'Plak \ (. * /) -> \ 1Pla'Ku / 2
• \ (. * /) Na'mezti \ (. * /) -> \ lNa'meztim / 2

In the (already described) steps g2p and wave-gen, phonemes are generated and then the acoustic signal is output.

The disadvantage is that the optimized phonemes or phoneme sequences are not used in the database. Instead, the graphemes are reshaped and the automatic converter (g2p) is used. This usually produces a lower quality in the pronunciation, as it knows only rules and only a few exceptions. Also, the step g2p requires computation time and makes the system unnecessarily sluggish.

• Graphem = Hauptplatz -> (reg-expr) Hauptplatzes -> (g2p) Aoptplases -> wave-gen.
• Korrekt wäre Aoptplazes. Es ergibt sich eine Abweichung an der markierten Stelle.

Here is an example:

• Graphem = Main Square -> (reg-expr) Main Square -> (g2p) Aoptplases -> wave-gen.
• Correct would be Aoptplazes. This results in a deviation at the marked point.

It is often considered acceptable when a speech synthesis incorrectly pronounces some words or only partially pronounces them correctly. But with navigation systems in particular, there is a high risk that some street and place names that are on the driver's daily driving route are pronounced very often, and the resulting incorrect pronunciation reduces the user experience.

In this context, the document discloses DE 100 42 942 A1 the use of phonetic transcriptions, which firstly involves a process of producing a phoneme from a grapheme and, secondly, the generated phoneme. Further, this document teaches the use of filler particles. The latter are used to generate phonemes or sequences of graphemes by storing parts of the graphemes (subwords) in a graph-based database and reassembling the corresponding part phonemes. For example, a "genitive" is considered as filler particle at the front partial word: "litigation" is broken down, for example, into the partial word "right", the filler particle "s" and the partial word "conflict". These three parts are individually translated into phonemes or phoneme sequences by means of a database and then reassembled. This document thus teaches how compound words are decomposed, parts of a piece are transcribed (converted) into phonemes or sequences of phonemes, and these are reassembled.

The publication DE 689 13 669 T2 generally deals with the identification of language groups as well as the rewriting of graphemes into phonemes.

" Message-to-Speech: "High Quality Speech Generation for Messaging and Dialog Systems", Peter Spyns et al., Concept on Speech Generation Systems Proceedings of a Workshop in Conjunction with the 35th Annual Meeting of the Association for Computational Linguistics, July 11, 1997 , Pages 11 to 16 discloses a message-to-speech system for spoken dialogue and message generation. The system uses prosody transplantation and specialized prosody models to obtain highly natural prosody for synthesized speech. The prosody transplantation is based on taking over the intonation and duration of speech from a human-language donor message into the phonetic transcription of the same message.

The publication DE 100 42 942 A1 discloses a method, apparatus, and computer program product for speech synthesis by grapheme-phoneme translation. Here, for a given word in a database containing phonetic transcriptions of words, it searches for subwords of the given word. If at least one partial word of the given word is found in the database, a phonetic transcription recorded in the database is selected for the partial word found. The given word has, in addition to the found subword, at least one further constituent that is not recorded in the database. This further component is phonetically transcribed by means of an OOV treatment and the phonetic transcription of the found subword and the phonetic transcription of the further component are combined.

Short outline

The present invention seeks a solution for at least one of the problems described above as well as for solutions to further problems. In particular, the user experience is to be increased when using navigation systems with voice-based route guidance.

According to the invention, a method and a navigation system according to the independent claims are provided. Developments are presented in the dependent claims.

The method includes the steps of storing a plurality of phoneme sequences in a first state in at least one database, selecting at least one of the plurality of phoneme sequences based on a control signal received as part of a route guidance, determining a desired second state of the at least one selected phoneme sequence on the basis of the selected one Context information, changing at least a portion of the selected phoneme sequence to translate the selected phoneme sequence to the specified second state, based on a policy, and converting the phoneme sequence in the second Condition into an acoustic signal for output by means of an acoustic output device.

Already existing phonemes are used and, for example, the case declined by means of replacement rules.

Furthermore, the phoneme sequence with the at least one changed subsequence is converted into an acoustic signal.

Preferably, therefore, no allophones are used, but phoneme sequences for the conversion from text to speech and vice versa. In one possible implementation, the phoneme alphabet used would be provided by the manufacturer of the TTS engine. This list of phonemes used is e.g. a specialization of the ASCII-based phonetic alphabet SAMPA. Advantageously, a subsequence of a phoneme sequence is identified and replaced (or such a subsequence appended to the phoneme sequence) such that a semantic component is irrelevant.

Preferably, the changing step may include attaching an attachment to the selected phoneme or phoneme sequence. In particular, the appendix may be the changing phoneme sequence and appended to the selected phoneme sequence. This can simplify the modification of the selected phoneme or phoneme sequence.

The received control signal may originate from a route calculation module of the navigation system. Furthermore, the navigation system can be arranged in a vehicle. The received control signal comprises a first identifier of a designation of a location determined by the navigation system and a second identifier of a maneuver instruction to be issued. In the latter case, the acoustic signal may comprise a designation of a location determined by the navigation system. If so, then the location determined may include at least one of the following: country name, city, street, place, and point-of-interest.

The storing step further comprises storing a plurality of predefined context information defining the second state. Furthermore, the selecting step further comprises selecting the at least one phoneme or the at least one phoneme sequence on the basis of the first identifier and selecting at least a predefined context information based on the second identifier. The determining step comprises determining the desired second state of the phoneme based on the selected context information. Finally, the method may further comprise outputting the acoustic signal by means of the acoustic output device, in which the designation is output as a phoneme sequence in the second state.

These additional (partially optional) features allow, for example, an optimal utilization of the already stored phonemes or phoneme sequences while optimally exploiting the system resources, since the decision on the second state alone from the context information determinable and the selected phoneme or the selected phoneme sequence directly into the second state is convertible. In other words, e.g. the case to be used is not already determined by the navigation system, but emerges from the sentence to be spoken, which is stored in the database.

One variant allows optimal declination, e.g. a street name using the already stored phonemes or phoneme sequences.

Preferably, the undeclared case may further be a nominative, while the declined case may be at least one of genitive, dative, accusative, vocative, locative, instrumental, ablative, allative, and prepi-positive. The part of the phoneme or phoneme sequence to be changed can be replaced by a regular expression. In the latter case, the attachment may be determined by a regular expression. This allows a further improved declination, e.g. street names using the regular expressions (which, for example, are themselves as a completed phoneme of a respective fall-ending, e.g., a spoken genitive-s). As described above, the phoneme sequence may be in the form of a noun nominative; however, advantageously, this does not require any linguistic intelligence that necessitates semantic treatment by morphemes.

The phonemes or phoneme sequences can each be present in a multiplicity of languages in the language database. The method may further include selecting one of the plurality of languages based on user input. This in turn allows optimal utilization of the already existing phonemes or phoneme sequences taking into account the desired language.

The multitude of languages may include Romance and Germanic languages, and the declined case may be one of genitive, dative, and accusative. This allows the use of e.g. German, French or English and an adaptation to the particular case based on the phoneme or phoneme sequences.

Alternatively, or in addition, the plurality of languages may include Slavic and Baltic languages. The declined case can here be one of genitive, dative, vocative and locative. Furthermore, the declined case may further include an instrumental and prepositional as far as the Slavic or Baltic language includes this case. This allows adaptation of the speech output system e.g. to the Polish, Czech or Slovak languages, taking into account that in those languages the accusative is different from the nominative (this is often not the case in German). Furthermore, the case locative (location dependency of the noun) is covered, which does not exist in German and can occur frequently in a navigation system.

Alternatively or additionally, the plurality of languages may include Oghuzian languages. Here, the declined case can be one of genitive, dative, and locative. The declined case may further include one of instrumental, ablative, and allative, as far as the Oghuzian language encompasses this case. This allows adaptation of the speech output system e.g. to the Turkish language, taking into account that in this language the accusative is different from the nominative. Furthermore, the case Ablativ (where?) And Allativ (where?) Are covered, which do not exist in German and are common in navigation systems, such. in type "Leave the x-road and turn into the y-lane".

Preferably, the phoneme or the phoneme sequence can represent a grammatical noun. The first state of the phoneme or phoneme sequence may be a singular of the noun, and the second state of the phoneme or phoneme sequence may be at least one of Dual, Paral, Trial, Quadral, Paukal, Plural, and Distributive of the noun. Alternatively or additionally, the phoneme or the phoneme sequence can represent a grammatical noun. The first state of the phoneme may be a masculine of the noun, and the second state of the phoneme or phoneme sequence of at least one feminine and neuter of the noun (or vice versa). This allows a possibly necessary adaptation of the noun also to the number and / or gender.

Preferably, a computer program product with program code sections is also provided for carrying out the method according to the invention when the computer program product is executed on one or more computer devices (for example a navigation system). The computer program product may be recorded on a computer readable recording medium.

Furthermore, a navigation system is provided for outputting an acoustic signal, wherein the navigation system comprises at least one database and an acoustic output device, and wherein the device comprises at least one processor which adapts the device to perform all the steps of the method presented here.

Brief description of the drawings

• Fig. 1 ein rein graphembasiertes Sprachausgabesystem;
• Fig. 2 ein rein phonem- bzw. phonemfolgenbasiertes Sprachausgabesystem;
• Fig. 3 ein mehrfachwandelndes Sprachausgabesystem;
• Fig. 4 ein erstes Prinzipschema eines Sprachausgabesystems gemäß einem Ausführungsbeispiel;
• Fig. 5 ein zweites Prinzipschema eines Navigationssystems gemäß einem Ausführungsbeispiel;
• Fig. 6 die Komponenten, die in einem Ausführungsbeispiel von einer Vorrichtung umfasst sind; und
• Fig. 7 ein Verfahren (und demgemäß die Interaktion zwischen den Komponenten) gemäß einem Ausführungsbeispiel.

The accompanying drawings show embodiments to which, however, the present invention is in no way limited. In the drawings, like reference numerals designate the same or similar functional blocks or steps. It should be noted that the presentation of individual functional blocks or steps does not preclude the possibility that the respective underlying functionality is to be implemented on several devices or in several steps. Show it:

• Fig. 1 a purely graph-based speech output system;
• Fig. 2 a purely phoneme-based speech output system;
• Fig. 3 a multi-conversion voice output system;
• Fig. 4 a first schematic diagram of a speech output system according to an embodiment;
• Fig. 5 a second schematic diagram of a navigation system according to an embodiment;
• Fig. 6 the components included in one embodiment of a device; and
• Fig. 7 a method (and hence the interaction between the components) according to one embodiment.

Detailed description

In the following description, for purposes of explanation but not limitation, specific details (such as individual signaling steps) are described to provide a thorough understanding of the technique presented herein. It will be apparent to those skilled in the art that the present technique may be practiced in other embodiments that depart from these specific details.

Further, those skilled in the art will appreciate that the following explained services, functions, and steps are made using software provided in combination with a microprocessor or using an Application Specific Integrated Circuit (ASIC) digital signal processor. DSP) or a general purpose computer. It should also be understood that while the following embodiments are described in the context of methods and apparatus, the technique presented herein may also be implemented in a computer program product as well as in a system including a computer processor and a memory coupled to the processor Memory contains one or more programs that perform or perform services, functions, and steps disclosed herein.

Fig. 4 shows a first schematic diagram of a system 200 according to an embodiment for implementing a navigation system. As in Fig. 4 2, the system 200 includes a database 2001 capable of storing both graphems and phoneme sequences in a first state (eg, nominative), an acoustic output device 2002, and a speech synthesis device 2003. The speech synthesis device 2003 may perform the function reg-exp Append or change a part of a selected phoneme or a selected phoneme sequence, for example, by a regular expression, so that the phoneme or phoneme sequence in a second state (eg, declined in the genitive) is present, and the wave-gene function already described for converting the phoneme or the phoneme sequence in the second state into an acoustic signal. The acoustic signal thus generated can in turn be output by means of the acoustic output device 2002.

The phonemes or phoneme sequences can be written in a format like Nuance-LH +, Star-Rec®, internationalephoneme or similar. be implemented. However, the present disclosure is not limited thereto.

• \ (.*/)' Plak \(.*/) --> \1Pla'Ku/2
• z.B.: 'Plak Pilzutz'kiego --> Pla'ku Pilzutz'kiego (polnisch)
• \ (.*/)Na'mezti \(.*/) --> \1Na'meztim/2
• z.B.: Na'meZti Mirr'u --> Na'meztim Mirr'u (tschechisch)

As an example of application, as in Fig. 4 is shown, a set of rules are defined, the rule is set up to apply the case of eg street and place names for the navigation system, which are stored eg in the nominative, directly to phonemes or phoneme sequences. In the step reg-exp, these rules, such as regular expressions, analyze phonemes or phoneme sequences and transform them into the appropriate case. Two suitable rules are for example:

• \ (. * /) 'Plak \ (. * /) -> \ 1Pla'Ku / 2
• eg: 'Plak Pilzutz'kiego ->Pla'kuPilzutz'kiego (Polish)
• \ (. * /) Na'mezti \ (. * /) -> \ 1Na'meztim / 2
• eg: Na'meZti Mirr'u ->Na'meztimMirr'u (Czech)

A regular expression (or regular expression) can be understood as a search pattern within a text string. Sometimes the combination of the search pattern and the replacement pattern is called a regular expression. For example, a regular expression can describe a phoneme attachment.

• Phonem bzw. Phonemfolge im Nominativ
• Sprache des Phonems bzw. der Phonemfolge
• benötigter Fall

Input parameters of a rule work based on regular expressions can be:

• Phoneme or phoneme sequence in the nominative
• Language of the phoneme or the phoneme sequence
• required case

• Phonem bzw. Phonemfolge im gewünschten Fall

Output parameter can be:

• Phoneme or phoneme sequence in the desired case

• Ein Regelwerk, z.B. als "Regular Expressions" (d.h. reguläre Ausdrücke) formuliert, wird hier für Phoneme eingesetzt.

Implementation can be:

• A set of rules, eg formulated as "regular expressions", is used here for phonemes.

In an exemplary implementation, the above-described changing subsequence exists as a "RegExp" (regular expression). A "RegExp" is a particularly memory-conserving variant of a (sub-) phoneme sequence, which is basically always pronounced the same way under different circumstances (ie even in the context of other phonemes or phoneme sequences does not or only insignificantly changes its pronunciation).

As an example of using a "RegExp" lies as in Fig. 4 shown in its ground state, for example, the phoneme sequence / Aoptplaz /. According to an external input (control signal), for example, the genitive of the phoneme sequence may be desired (this does not mean that a semantic component would be involved, eg a simple if-then implementation). As a result, the phoneme / es / (here: a genitive-s) is appended to the phoneme sequence; since the genitive-s is always pronounced the same at the end of the word, a "RegExp" can be used instead of a "full-blown" phoneme or the phoneme sequence. The new phoneme sequence / Aoptplazes / is now converted into an acoustic signal, which can be carried out particularly efficiently and memory-conserving when using a "RegExp".

Fig. 5 shows a second schematic diagram of a system 200 for navigation purposes, which is based on the system according to Fig. 4 based. As in Fig. 5 is shown, the navigation system 200 includes the database 2001, both grapheme and phonemes or phoneme sequences in a first state (eg nominative) and optional context information (eg a ready-made sentence as a grapheme, in which only the phoneme or phoneme sequence can be fitted in the appropriate case), the acoustic output device 2002 and the speech synthesis device 2003 and a navigation device 2004 with a route guidance module. The speech synthesis device 2003 may include the functions g2p, reg-exp and wave-gen described above. The acoustic signal thus generated can in turn be output by means of the acoustic output device 2002. Further, the system 200 may still generate and / or process signals 2005, 2006, and 2007, which are described in more detail below.

Fig. 6 shows the components included in one embodiment of the device (speech synthesis device) 2003. The (speech synthesis) device 2003 comprises a core functionality 20031, which may be, for example, as at least one CPU (central processing unit) or microprocessor, as dedicated Circuit (with the implementation variants described above) or implemented as a software module. Furthermore, the device 2003 comprises a memory 20032, a transmitter 20033 and a receiver 20034, which may be used to communicate the device 2003 with another device (eg the database 2001, the navigation system 2004 or the acoustic output device 2002) or with the user. Furthermore, the device 2003 comprises a selector 20035, a determiner 20036, a modifier 20037 (which optionally may in turn comprise a trailer 20038), and a converter 20039. Likewise, the database 2001 comprises a core functionality 20011, a memory 20012, a transmitter 20013, and a Receiver 20014, the acoustic voice output device 2002 comprises a core functionality 20021, a memory 20022, a transmitter 20023 and a receiver 20024, and the navigation device 2004 comprises a core functionality 20041, a memory 20042, a transmitter 20043 and a receiver 20044.

The core functionality 20041 of the navigation device 2004 may include a software-based route calculation module. This route calculation module may be configured to generate, transmit (via the transmitter 20043), receive (via the receiver 20044), and process various control signals 2005, 2006, 2007. Thus, the control signal 2005 "Prepare Prompt" can communicate one or more identifiers with regard to the phonemes or phoneme sequences to be selected from the database 2001 to the speech synthesis device 2003. The selection based thereon and optionally speech synthesis can be confirmed by means of the control signal 2006 "Preparation Finished" with respect to the navigation device 2004, whereupon the speech synthesis device 2003 prompts the speech synthesis device 2003 to output the synthesized acoustic signal by means of the acoustic output device 2002 by means of the control signal 2007 "Prompt Now".

As in Fig. 6 indicated by the dashed extension of the core functionalities 200x1 (where x = 1, 2, 3, and / or 4), all of the aforementioned devices illustrated within the dashed lines may be implemented both as stand-alone devices and as sub-functionalities of the core functionality be. All of the aforementioned devices that overlap the dashed area may be driven by the core functionality 200x1 or may provide information to the core functionality 200x1.

The core functionalities 200x1 may be configured, for example, by software resident in the memories 200x2 to process various data inputs and to control the functions of the memory 200x2, the transmitter 200x3 and the receiver 200x4 (and the selector 20035, the detector 20036, the modifier 20037, the trailer 20038 and the converter 20039 of the speech synthesizer 2003). The memory 200x2 may be for storing code portions for carrying out the methods according to the aspects described above when running on the core functionality 200x2.

It should be noted that transmitter 200x3 and receiver 200x4 may alternatively be provided as a one-piece transceiver, as in FIG Fig. 6 is shown. It should also be noted that the transceivers may be implemented as: physical transceivers for transceiving via an air interface (eg between the navigation system 2004 and a traffic guidance system not shown), as traffic-routing functional units / interfaces between network elements (eg for transmission / Receiving data packets between the speech synthesis device 2003 and the navigation system 2004 and the database 2001) as functionality for writing / reading information to / from a given memory area (eg between the speech synthesis device 2003 and the database 2001 or the navigation device 2004, if these in a network unit) or as any suitable combination of the arrangements described above. At least one of the selector 20035, determiner 20036, modifier 20037, hitch 20038 and converter 20039 (the speech synthesizer 2003) or the respective executing functionalities may also be implemented as a chipset, module or subunit.

Fig. 7 shows an embodiment of a method for outputting an acoustic signal by means of a navigation system (eg according to FIG Fig. 5 ). In the in Fig. 7 The signal flow diagram shown, the signal flow between elements in the horizontal direction is given, whereas temporal aspects between the signaling in the vertical arrangement of the signal flow sequence and in the sequence numbers are reproduced. It should be noted that the in Fig. 7 shown temporal aspects none of the steps shown on the in Fig. 7 Set the step sequence shown. This relates in particular to process steps which are functionally disjunctive from one another. For example, step S1 (storing phonemes or phoneme sequences) is shown substantially simultaneously with step S2 (selecting at least one phoneme or phoneme sequence). However, this does not exclude that the phonemes or phoneme sequences are stored in step S1 long before the device according to the invention is put into operation.

With reference to the signal flow plan according to Fig. 7 (which together with the speech synthesizer 2003, the database 2001 and the navigation device 2004 in Fig. 5 and 6 to be read), in step S1, the memory 20012 of the database 2001 stores a plurality of phoneme sequences in a first state in the at least one database to bind to the at least one database. Optionally, a plurality of predefined context information defining the second state may be stored in the memory 20012 of the database 2001. As in Fig. 5 For example, this context information may be a predefined sentence (eg stored as a grapheme), where the second state (eg the desired case) results from embedding the phoneme or the phoneme sequence in the sentence structure.

In step S2, the selector 20035 performs selection of at least one of the plurality of phoneme sequences on the basis of a control signal received in the course of route guidance. The control signal may originate from a route calculation module (not shown) of the navigation device 2004. As indicated by the signal 2005 in Fig. 5 In addition, the received control signal may comprise a first identifier (ID_Main) of a designation of a location determined by the navigation device 2004 and a second identifier (ID_In_200m_right) of a route guidance or maneuver indication to be output. The first identifier can be used to select the phoneme or the phoneme sequence (here: 'Aoptplaz) (in the simplest case a simple indexing can be used). The second identifier can be used to call a corresponding context information, so that - as described above - the desired second state (eg the desired case) results by embedding the phoneme or the phoneme sequence in the sentence structure. This is advantageous since the navigation device 2004 only has to send a simple message with two identifiers to the speech synthesis device 2003, which contributes to a streamlined signaling (which is advantageous, for example, if the navigation device 2004 does not co-locate with the database 2001 or device 2003) is, but for example, a tight (and possibly shared) bus for data transmission between the above devices is provided).

In step S3, the determination means 20036 of the speech synthesis apparatus 2003 determines a desired second state of the at least one selected one Phones or the phoneme sequence on the basis of the received control signal. As already described, the desired second state of the phoneme or the phoneme sequence can be determined on the basis of the selected context information, in particular, the second identifier can be used to call a corresponding context information, so that the desired second state (eg the desired case) by Embedding the phoneme or phoneme sequence can be determined in the sentence structure. In this context, the phoneme or the phoneme sequence can represent a grammatical noun. Further, the first state of the phoneme or phoneme sequence may be an undefined case (eg nominative) of the noun and the second state of the phoneme or phoneme sequence may be a declined case (eg genitive, dative, accusative, vocative, locative, instrumental, ablative, allative and / or prepositive) of the noun.

In step S4, the changer 20037 of the speech synthesizer 2003 changes at least part of the selected phoneme or phoneme sequence to place the selected phoneme or phoneme sequence in the specified second state, based on a set of rules. In this case, the appendix 20038 of device 2003 may append an appendix to the selected phoneme or phoneme sequence. In particular, this appendix (but also the part to be changed) can be replaced by a regular expression (eg an always same case ending, deposited as a phoneme or phoneme sequence, such as the genitive-s already described or the phoneme or phoneme sequence of the suffix "it be formed in phoneme sequences such as' Aoptplaz). The change may alter at least a portion of the phoneme sequence of the undefined noun to produce the declined noun, or may append an appendix to the noun to add the declined noun, as just described.

The set of rules described above can be constructed, for example, as follows: rule Target case Filter / pattern replacement #1 Genitive \ (. * \) Plaz \ 1 plazes # 2 Genitive \ (. * \) DOAF \ 1 doafes # 3 majority \ (. \) Stäle \ 1 steal

In step S5, the converting means 20039 converts the phoneme in the second state into an acoustic signal for output from the acoustic output device 2002. The acoustic signal may include a designation of a location obtained by the navigation device 2004. In this case, the location determined may include at least one of the following: country name, city, street, place, and point-of-interest.

In an optional step S6, the speech synthesis device 2003 outputs the acoustic signal via the transmitter 20033 to the acoustic output device 2002, which acoustically reproduces the designation as the phoneme or the phoneme sequence in the second state. In the simplest case, the speech output device is a loudspeaker that reproduces the acoustic signal (as an output of the wave gene function). However, the acoustic output device 2003 may also be more complex, e.g. can the speech synthesis be content with the generation of the phonemes or phoneme sequences, in which case the function wave gen can be outsourced to the acoustic output device 2002.

Another example is in Fig. 5 signals 2006 "Preparation Finished" and 2007 "Prompt Now" shown. The preparation of a voice output can take a predetermined time (depending on the system, for example, 1 to 5 seconds). Thus, the navigation device 2004 may send the control signal 2005 earlier (preferably plus a temporal safety allowance) to the device 2003 by the predetermined time to already prepare the phoneme to be outputted or the phoneme sequence to be output. After receiving the control signal 2005, steps S2 (Select) to S4 (Change) (or to S5, Convert, depending on the equipment of the acoustic output device 2002) would then already take place and a successful completion of the phoneme sequence by signal 2006 could the navigation system 2004 be reported back. If the acoustic output is actually needed, then the navigation device 2004 outputs the signal 2007, whereupon steps S5 and S6, output (or only S6) can then be performed. This procedure is advantageous in order to adapt the presented technology, for example, to real-time conditions.

As already mentioned, the phonemes or phoneme sequences may be in a variety of languages and accordingly may be selected based on user input. In the case of Romance and Germanic languages, the declined case can be one of genitive, dative, and accusative. In the case of Slavic and Baltic languages, the declined case may be one of genitive, dative, vocative, and locative (instrumental and pre-positive, if any). In the case of Oghuzian languages, the declined case may be one of genitive, dative, and locative (instrumental, ablative, and allative, if any).

Finally, the first state of the phoneme or phoneme sequence may still be a singular of the noun, while the second state of the phoneme may be at least one of Dual, Paral, Trial, Quadral, Paukal, Plural, and Distributive of the noun. Furthermore, the first state of the phoneme or phoneme sequence may be a masculine of the noun, the second state of the phoneme or phoneme sequence may be at least one of feminine and neuter of the noun.

As has been apparent from the above-described embodiments, the technique presented leads to one or more advantages.

In the case of navigation systems permanently installed in motor vehicles, suppliers can be pre-selected for a (navigation) project and do not need to be changed so that the software version is rarely changed after delivery (as software retrofit is currently only available in one) Workshop can be carried out). Thus, the implementation based on phonemes or phoneme sequences is no problem.

The circumvention or non-use of graphemes, as used in the prior art, is an advantage of some of the embodiments.

The phoneme formats are often language dependent. Therefore, the language can be chosen before applying the algorithm. In the case of a permanent installation, the system-level language setting is performed for all affected vehicle components. Thus, only one language applies at a time, and the system presented here can make the appropriate settings. In some cases, the system is bilingual: names of places, streets, etc. are rendered in the local language, while all other sentences and phrases are pronounced in the system language. Country-specific variants are also produced in conventional (navigation) projects in the automotive industry, so that the required language-dependent regular expressions for various markets can be created in the course of the project and only the required part must be provided upon delivery.

Often, not all languages (including their idiosyncrasies) are known before completion of the navigation system, and so quality losses are only in the test on. This late detection usually leads to missed deadlines or great willingness to compromise to hold appointments. The proposed technique thus helps in the project management, since they from the outset the o.g. Error avoids.

Claims (11)

1. A method for output of an acoustic signal by means of a navigation system (200), wherein the navigation system comprises at least one database (2001) and an acoustic output means (2002), comprising the steps of:

   storing (S1) of:

   - a plurality of phoneme sequences in a first state and

   - a plurality of predefined context information, which define a second state, in the at least one database (2001), wherein:

   - the phoneme sequence represents a grammatical noun,

   - the first state of the phoneme sequence is a non-declined case of the noun, and

   - the second state of the phoneme sequence is a declined case of the noun;

   selecting (S2) at least one of the plurality of phoneme sequences based on a control signal received within the framework of a route guidance (2005), wherein the received control signal comprises a first identifier of a designation of a location found by the navigation system (200) and a second identifier of a maneuvering hint to be output, and wherein the step of selecting further comprises:

   - selecting the at least one phoneme sequence based on the first identifier;

   - selecting at least one of the predefined context information based on the second identifier;

   determining (S3) a desired second state of the at least one selected phoneme sequence based on the selected context information;

   changing (S4) at least one part of the selected phoneme sequence in order to transit the selected phoneme sequence into the specified second state based on a set of rules, wherein the part to be changed of the phoneme sequence is replaced by means of a regular expression; and

   converting (S5) the phoneme sequence in the second state into an acoustic signal for output by means of the acoustic output means (2002).
2. The method according to claim 1, wherein
   the step of changing comprises annexing an annex to the selected phoneme sequence, wherein in particular the annex is a changing phoneme sequence and is annexed to the selected phoneme sequence.
3. The method according to at least one of the preceding claims, wherein
   the received control signal stems from a route calculation module.
4. The method according to at least one of the preceding claims, wherein
   the navigation system (200) is arranged in a vehicle.
5. The method according to any one of the preceding claims, wherein
   the acoustic signal comprises a to-be-output designation of a location found by the navigation system (200).
6. The method according to claim 1 or 5, wherein
   the found location comprises at least one of the following indications: country name, city, street, square and point of interest.
7. The method according to claims 1, further comprising:

   outputting (S6) the acoustic signal by means of the acoustic output means.
8. The method according to claim 2, wherein the annex is determined by the regular expression.
9. A computer program product having program code portions for performing the method according to at least one of the preceding claims, when the computer program product is executed on one or more computing means.
10. The computer program product according to claim 9, which is recorded on a computer-readable recording medium.
11. A navigation system (200) for output of an acoustic signal, wherein the navigation system (200) comprises at least one database (2001), an acoustic output means (2002) and at least one processor (20031), which processor adapts the navigation system to perform all the steps of a method according to at least one of claims 1 to 8.

Images