DE102019216078B3 - Method and apparatus for identifying instructions in speech recorded in a vehicle - Google Patents

Abstract

A computer implemented method for identifying instructions in speech recorded in a vehicle is provided. The method includes providing a list of instruction words and a list of topic words relating to topics of conversation. A unit of speech is recorded in a vehicle and compared with the instruction words and the topic words in the vehicle. When a topic word is identified, the topic word is added to a current topic list. When an instruction word is identified, information relating to the recorded speech unit and the current topic list is transmitted to a remote server. In the remote server, an instruction is identified based on the transmitted information and instruction data is generated based on the identified instruction. The instruction data are transmitted from the server to the vehicle and processed in the vehicle.

Description

Field of invention

The invention relates to a computer-implemented method, a device and a system for identifying instructions in speech recorded in a vehicle.

Background of the invention

Many modern vehicles have in-vehicle voice recognition systems that enable the driver and / or other passengers to communicate with the vehicle systems using voice instructions. To recognize these voice instructions, a voice recorder can continuously record the voices within the vehicle.

In the US application US 2013/0 103 405 A1 describes an operating system that includes a requested task estimation unit that estimates an applicant for a requested task based on information contained in an utterance of a first speaker.

The US filing US 2016/0 085 854 A1 describes how a computing device that is in conversation with a user is enhanced by identifying and managing one or more topics. With the techniques described, the computing device can concentrate on one or more topics, switch back and forth between topics and / or introduce new topics.

In the US application US 2018/0 233 143 A1 describes a dialog system that receives data via speech / text signals, updates probabilistic system states, identifies a relevant category via an identification model, excludes irrelevant categories and determines an output action (e.g. text or synthesized speech).

It can be advantageous to carry out the identification of voice instructions in a remote server rather than locally in the vehicle. Indeed, a remote server enables more processing power to be used and easier / faster software and hardware updates.

However, due to privacy concerns or limited communication bandwidth available between the vehicle and the remote server, recorded voices are not continuously transmitted to such a remote server. Only when an instruction word has been identified in the vehicle is information relating to a corresponding recorded voice instruction transmitted to the remote server. This information is processed remotely to identify the complete voice instruction and instruction data is remotely generated based on the identified instruction. This instruction data is transmitted to the vehicle which then processes the instruction data, i.e. H. the vehicle carries out the identified voice instruction.

Object of the invention

The aim of the invention is to improve the recognition of instructions in speech recorded in a vehicle.

Summary of the invention

The object of the invention is achieved by a method according to claim 1, an apparatus according to claim 10 and a system according to claim 12.

• - Bereitstellen einer Liste mit Anweisungswörtern;
• - Bereitstellen einer Liste mit Themenwörtern bezüglich Gesprächsthemen;
• - Aufzeichnen einer Spracheinheit in einem Fahrzeug;
• - Vergleichen der aufgezeichneten Spracheinheit mit den Anweisungswörtern und den Themenwörtern in dem Fahrzeug;
• - wenn ein Themenwort unter Verwendung des Vergleichs in der Spracheinheit identifiziert wird, Hinzufügen des Themenworts zu einer aktuellen Themenliste;
• - wenn ein Anweisungswort unter Verwendung des Vergleichs in der Spracheinheit identifiziert wird, Übertragen von Informationen bezüglich der aufgezeichneten Spracheinheit und der aktuellen Themenliste vom Fahrzeug zu einem Fernserver;
• - Identifizieren einer Anweisung basierend auf den Informationen in dem Fernserver;
• - Erzeugen von Anweisungsdaten basierend auf der identifizierten Anweisung in dem Fernserver und Übertragen der Anweisungsdaten vom Server zum Fahrzeug; und
• - Verarbeiten der Anweisungsdaten in dem Fahrzeug.

A computer implemented method for identifying instructions in voice recorded in a vehicle is provided, comprising the steps of:

• - providing a list of instruction words;
• - providing a list of topic words relating to topics of conversation;
• - recording a speech unit in a vehicle;
• - comparing the recorded speech unit with the instruction words and the subject words in the vehicle;
• - when a topic word is identified using the comparison in the speech unit, adding the topic word to a current topic list;
• if an instruction word is identified using the comparison in the speech unit, transmitting information relating to the recorded speech unit and the current topic list from the vehicle to a remote server;
• - identifying an instruction based on the information in the remote server;
• - generating instruction data based on the identified instruction in the remote server and transmitting the instruction data from the server to the vehicle; and
• - processing the instruction data in the vehicle.

The language recorded in the vehicle can be the words and phrases spoken by the driver and / or the other passengers in the vehicle. The vehicle can be a car, motorcycle, pickup truck, or truck. It can also be a bicycle or a moped.

The instructions in the recorded language may relate to the operation of the vehicle and / or the infotainment system of the vehicle. For example: "Open the left rear window"; "Play the next song", "What is the building on the left?", "Show a photo of the Eiffel Tower", "Calculate a route to Pierre's". The instructions can have a hotword or an instruction pointer, such as "hello assistant" or just "assistant". This hotword can indicate that a sentence started or ended with an instruction. For example: “Assistant: calculate a route to Pierre's” or “What is the building on the left, assistant?”.

An instruction word can be part of an instruction or can (i) be an instruction such as "open", "show" or "calculate"; (ii) a question word such as "what" or "how"; (iii) a hotword as described above; or any combination of (i), (ii) and / or (iii).

A topic of conversation refers to the topic of a conversation, dialogue or monologue inside the vehicle. Examples of the topics of conversation are “Travel”, “Nutrition”, “Environment”, “Movies” and “Entertainment”.

A topic word is a word that relates to at least one specific topic of conversation; H. it can indicate a topic of a conversation. For example, topic words such as “hunger” and “dinner” relate to the topic of conversation “nutrition” and can be assigned to it. Topic words such as "building" and "tree" relate to the topic of conversation "environment" and can be assigned to it. A topic word “noise” can refer to both the topic of conversation “entertainment” and the topic of conversation “environment”. The list of topic words can include topic words and the assigned topics of conversation.

The phrase “a recorded unit of speech” refers to a recording of speech over a period of time. For example, a recorded speech unit can be 3 seconds of recorded speech.

When a topic word is identified in a recorded speech unit (locally, i.e. in the vehicle), the identified topic word is added to a current topic list. During the conversation in the vehicle, many topic words can be added to the current topic list.

When an instruction word is identified in a recorded speech unit (locally, i.e. in the vehicle), information is transmitted to the remote server. This information can include information relating to the recorded speech unit which can represent the complete recorded speech unit and / or certain characteristics of the recorded speech unit.

The information transmitted to the remote server can also include information relating to the current topic list with the identified topic words. This information may represent the current topic list itself and / or the topic of conversation to which most of the topic words have been assigned.

The remote server is a computer server that is located outside the vehicle, for example in a central computer or data center. In the remote server, an instruction is identified based on the information relating to the recorded language unit and the current topic list with identified topic words.

Based on the identified instruction, instruction data is generated in the remote server and transmitted back to the vehicle. The instruction data may include the identified instruction.

After the vehicle receives the instruction data, it can process the instruction data. To do this, it can execute the instruction spoken inside the vehicle and recorded in the recorded speech unit.

One advantage can be that the remote server receives additional information about the topic of conversation in the vehicle and this improves the identification of instructions in the recorded speech unit.

Since only information relating to the identified topic words is transmitted to the remote server, no data protection-sensitive information is transmitted to the outside of the vehicle, i. H. shared with the outside world.

Another advantage is that instead of the entire conversation, only information relating to the identified topic words is transmitted to the remote server. Therefore, the requirements for the communication connection (e.g. a 4G / 5G or WiFi connection) between the vehicle and the remote server are limited.

In one embodiment of the method according to the invention, the current topic list has topics of conversation, the topic words being assigned to the topics of conversation. In another embodiment, the information relating to the current topic list is information relating to the topic of conversation to which the most topic words have been assigned.

One advantage of this embodiment is that instead of transmitting information relating to the entire conversation or the current list of topics, only information relating to a topic of conversation (for example the topic of conversation itself) is transmitted to the remote server. This means that even less information is transmitted to the outside of the vehicle.

In one embodiment of the method according to the invention, each added topic word is assigned a time stamp which indicates the time at which the topic word was added to the current topic list.

The subject of a conversation can change during the conversation. A conversation can end and a new one can start at a later time. The longer it has been since a topic word was identified, the less indicative it can be of the topic of the current conversation. Therefore, a timestamp can help assess the information in the current topic list.

In a further embodiment of the method according to the invention, an added topic word is removed from the current topic list if a time difference between a current time and its associated time stamp is longer than a predefined period of time.

An advantage of this embodiment is that topic words that have previously been identified and are unlikely to indicate the current topic of conversation are removed from the current topic list.

In another embodiment of the method according to the invention, all added topic words are removed from the current topic list if a time difference between a current time and a time stamp of an added keyword that was added last is longer than a predefined period of time.

An advantage of this embodiment is that topic words identified during a previous conversation are removed from the current topic list.

In one embodiment of the method according to the invention, the list of subject words is generated in the vehicle. The list of subject words can be generated by the vehicle or by a user of the vehicle. An advantage of this embodiment is that the list is not generated in the remote server, which limits the amount of information that is transmitted between the remote server and the vehicle.

In one embodiment of the method according to the invention, the topic words are assigned to topics of conversation based on rules, and the rules are preferably generated in the vehicle. An advantage of this embodiment is that the rules are not generated in the remote server, which limits the amount of information that is transmitted between the remote server and the vehicle.

In addition, a computer program is provided, the computer program being designed to carry out the method as explained in this document or according to any one of the claims, while the computer program is being executed on a computer.

In addition, a computer readable storage medium is provided which stores a set of code executable by a control unit which, when executed in an apparatus on a processing system or in a digital signal processor, performs the method steps as set out in this document or according to any of the claims.

• - ein Speicherungsmedium mit einer Liste mit Anweisungswörtern, einer Liste mit Themenwörtern bezüglich Gesprächsthemen und einer aktuellen Themenliste;
• - ein Sprachaufzeichnungsgerät, das zum Aufzeichnen einer Spracheinheit eingerichtet ist;
• - einen Sendeempfänger, der zum Austauschen von Informationen zwischen der Vorrichtung und einem Fernserver eingerichtet ist; und
• - einen Prozessor, der eingerichtet ist zum:
  1. (i) Vergleichen der aufgezeichneten Spracheinheit mit den Anweisungswörtern und den Themenwörtern;
  2. (ii) wenn ein Themenwort unter Verwendung des Vergleichs in der Spracheinheit identifiziert wird, Hinzufügen des Themenworts zu der auf dem Speicherungsmedium gespeicherten aktuellen Themenliste;
  3. (iii) wenn ein Anweisungswort unter Verwendung des Vergleichs in der Spracheinheit identifiziert wird, Anweisen des Sendeempfängers, Informationen bezüglich der aufgezeichneten Spracheinheit und der aktuellen Themenliste zu dem Fernserver zu übertragen;
  4. (iv) Verarbeiten von Anweisungsdaten, die durch den Sendeempfänger vom Fernserver empfangen wurden und in dem Fernserver basierend auf den Informationen erzeugt wurden.

An apparatus for identifying instructions in speech recorded in a vehicle further comprises:

• a storage medium with a list of instruction words, a list of topic words relating to topics of conversation and a current list of topics;
• a voice recording device which is set up to record a unit of speech;
• a transceiver configured to exchange information between the device and a remote server; and
• - a processor that is set up to:
  1. (i) comparing the recorded unit of speech with the instruction words and the topic words;
  2. (ii) if a topic word is identified using the comparison in the speech unit, adding the topic word to the current topic list stored on the storage medium;
  3. (iii) if an instruction word is identified using the comparison in the speech unit, instructing the transceiver to transmit information regarding the recorded speech unit and the current topic list to the remote server;
  4. (iv) processing instruction data received by the transceiver from the remote server and generated in the remote server based on the information.

The device can be in the infotainment system or a navigation system of a vehicle. The storage medium can be ROM or RAM computer memory. The processor can be the CPU of the infortainment or navigation system.

A vehicle with a device as described in this document or according to claim 10 is provided.

• - eine Vorrichtung wie in diesem Dokument oder gemäß Anspruch 10 beschrieben, oder ein Fahrzeug wie in diesem Dokument oder gemäß Anspruch 11 beschrieben; und
• - einen Fernserver, der eingerichtet ist zum:
  • (i) Empfangen von Informationen bezüglich der aufgezeichneten Spracheinheit und der aktuellen Themenliste;
  • (ii) Identifizieren einer Anweisung basierend auf den Informationen in dem Fernserver;
  • (iv) Erzeugen von Anweisungsdaten basierend auf der identifizierten Anweisung; und
  • (v) Übertragen der Anweisungsdaten zu dem Fahrzeug.

And a system for identifying instructions in recorded speech in a vehicle is provided, comprising:

• a device as described in this document or according to claim 10, or a vehicle as described in this document or according to claim 11; and
• - a remote server that is set up to:
  • (i) receiving information regarding the recorded speech unit and the current topic list;
  • (ii) identifying an instruction based on the information in the remote server;
  • (iv) generating instruction data based on the identified instruction; and
  • (v) transmitting the instruction data to the vehicle.

It can be understood that (i) the operation of this device, this vehicle and / or this system, (ii) the associated advantages and (iii) further embodiments of this device are in accordance with the corresponding method steps as in this document.

Figure list

• 1 zeigt schematisch eine Vorrichtung, ein Fahrzeug und ein System gemäß einer Ausführungsform der Erfindung;
• 2 zeigt schematisch eine Vorrichtung gemäß einer Ausführungsform der Erfindung; und
• 3 zeigt schematisch ein Verfahren gemäß einer Ausführungsform der Erfindung.

For a more complete understanding of the present invention, reference is made to the following description, taken in conjunction with the accompanying drawings, in which:

• 1 shows schematically an apparatus, a vehicle and a system according to an embodiment of the invention;
• 2 shows schematically an apparatus according to an embodiment of the invention; and
• 3 shows schematically a method according to an embodiment of the invention.

Detailed description of the invention

1 shows schematically an embodiment of a device 120 for identifying instructions in a vehicle 110 recorded speech.

In 1 is also a remote server 130 pictured. The remote server 130 and the vehicle 110 or the device 120 are in the system 100 for identifying instructions in speech recorded in a vehicle.

A communication link 180 between the remote server 130 and the device 120 allows data or information to be transferred between the two. The communication link 180 can be any wireless communication link, such as 4 / 5G, WiFi, or vehicle-to-X based communication links.

As in 2 shown, the device 120 a storage medium 210 with a list 220 with instruction words, a list 230 with topic words relating to topics of conversation and a current list of topics 270 with identified topic words. The device 120 further comprises a voice recorder 240 arranged to record a unit of speech, and a transceiver 250 that is used to exchange information between the device 120 and the remote server 130 is set up. The device 120 further comprises a processor 260 on.

In 3 Figure 3 is a schematic overview of a process 300 mapped according to an embodiment of the invention. The procedure 300 instructs the steps 310-390 on. The steps 330-360 are in the device 120 inside the vehicle 110 executed and the steps 370 and 380 are in the remote server 130 executed. The steps 310 and or 320 can in the remote server 130 , in the device 120 inside the vehicle 110 , during the manufacture of the device 120 are executed. This is indicated by the dashed lines in 2 specified.

In step 310 a list of instruction words is provided. The list 220 can be made by the manufacturer of the device 120 or the vehicle 110 and can be generated on the storage medium 210 the device 120 be saved. As described above, an example of such an instruction word is “calculate”. The list 220 can through the device 120 from the remote server 130 using the transceiver 250 in the device 120 be received.

In step 320 a list of topic words related to topics of conversation is provided. The topics of conversation can be provided by the manufacturer of the device 120 or the vehicle 110 defined or generated. As described above, an example of such a topic of conversation is “nutrition”. The topics of conversation can be through the device 120 using the transceiver 250 in the device 120 from the remote server 130 be received.

The list 230 with subject words relating to these topics of conversation can be made by the user of the device 120 or by the manufacturer of the device 120 or the vehicle 110 be generated. The list 230 can be on the storage medium 210 the device 120 get saved. Examples of topic words relating to the topic of "nutrition" can be the words "hunger", "dinner" or "eat".

In step 330 becomes a unit of speech by the voice recorder 240 in the vehicle 110 recorded. The voice recorder 240 can be set up for continuous recording of the speech or voices in the vehicle in speech units with a certain length of time, for example 2 seconds.
As an example, a first recorded speech unit may have voices that say, "I'm hungry,""Do you want to go out for dinner?" says. The recorded speech unit can be stored on the storage medium 210 get saved.

In step 340 becomes the recorded speech unit with the stored instruction words and the stored subject words in the vehicle 110 compared, for example by the processing unit 260 the device 120 . In one or more embodiments, information related to the recorded speech unit is compared with information related to the stored instructional words and topic words.

In step 350 when a topic word is identified in the speech unit using the comparison, the identified topic word becomes, for example, the processing unit 260 to the current list of topics 270 added. The current list of topics 270 may have the topics of conversation to which the identified topic words were assigned.

Following the previous example, the topic words "hungry" and "dinner" may be identified and possibly related to the current topic list 270 with the assigned topic of conversation “Nutrition”.

In step 360 when an instruction word using the comparison in the language unit (for example by the processing unit 260 ) is identified, information relating to the recorded speech unit and the current topic list is obtained from the vehicle 110 to the remote server 130 transfer. This information can be provided by the transceiver 250 using the communication link 180 be transmitted.

In the example, the instruction word “calculate” may have been used by the processing unit 260 identified in the second recorded speech unit. Then information related to the second recorded speech unit, for example the recorded voice “Assistant, calculate a route to Pierre's” himself, becomes the remote server 130 transfer. This information can be provided by the transceiver 250 using the communication link 180 be transmitted.

Further, information regarding the list of identified topic words "hungry" and "dinner" (for example these words) and / or the topic of conversation with most of the assigned identified topic words added (in this case, for example, the topic of conversation "diet") is transferred to the remote server 130 transfer. This information can be provided by the transceiver 250 using the communication link 180 be transmitted.

The information can be provided by the remote server 130 be received. In step 370 the instruction is identified based on the information transmitted. In the example, the instruction “Assistant, calculate a route to Pierre's” can be identified, and in particular the meaning of the word “Pierre's” in terms of the topic of conversation being “Diet” can be identified. In fact, the route to be calculated can be a route to a restaurant called "Pierre's".

In step 380 become instruction data in the remote server 130 generated based on the identified instruction. The instruction data is from the remote server 130 to the device 120 in the vehicle 180 transfer. The instruction data can correspond to the identified instruction. In the case of the example, the instruction data can be the device 120 in the vehicle 110 instructs to calculate a route to the address of the restaurant "Pierre's". The instruction data can therefore include the GPS coordinates of the restaurant obtained by the remote server 130 were determined. The instruction data can be sent by the transceiver 250 using the communication link 180 be received.

In step 390 becomes the instruction data in the device 120 (or more precisely by the processing unit 260 ) in the vehicle 110 processed or executed. In the example, the device 120 follow the instructions and calculate a route to the GPS coordinates of the restaurant "Pierre's".

Furthermore, there may be the case that a third recorded speech unit can have voices that say: “What's the name of Woody Allen's film again?” And “I saw it in the cinema”. The topic words "film" and "cinema" can then be identified and added to the current topic list. These topic words may have been assigned to the topic of conversation “Movies”. A fourth recorded speech unit may have a voice saying "Assistant, play Midnight in Paris".

If information about the current topic list is now transmitted to the remote server, this information can now clearly indicate whether the topic of conversation was "nutrition" or "cinema", since both topic words have been added to the current topic list.

Thus, in one embodiment, each added topic word may be assigned a time stamp indicating the time the topic word was added to the current topic list. In this case, topic words added later can specify the topic of the conversation. Following the example, information about the current list of topics will then indicate that the topic of conversation is more likely “cinema” than “nutrition”.

Another solution is to remove an added topic word from the current topic list when a time difference between a current time and its associated time stamp is longer than a predefined period of time. In the example, the topic words “hunger” and “dinner” may have been removed from the current topic list so that the information about the current topic list will indicate that the topic of conversation is “movies”.

The remote server may use the current topic information indicating that the topic of conversation is "cinema" to determine that the phrase "Midnight in Paris" refers to a movie, not a song. The instruction data, which the remote server can transmit back to the vehicle, can therefore have a pointer to a download location of the film "Midnight in Paris".

In one or more embodiments, all added topic words are removed from the current topic list when a time difference between a current time and a timestamp of an added keyword that was last added is longer than a predefined time period.

It can be understood that if no new topic words have been added to the current topic list during a period of time, it is likely that the conversation has temporarily stopped. It can therefore be advantageous to remove all topic words you have added from the current topic list before starting a new conversation with a likely new topic.

In general, the removal of topic words from the current topic list can be triggered by the identification of an instruction word. In this case, the topic words that may be less relevant are removed from the current topic list before information about the current topic list is transmitted to the server.

Alternatively, the functions of the remote server 130 (also) in the device 120 implemented. In this case, the procedure for identifying instructions can be carried out entirely in the vehicle 110 run and the connection 180 not necessary. The information regarding the recorded language unit and the current topic list would then not be sent to the remote server 130 but would be in the device 120 are processed. That is, an instruction appears in the information in the device 120 identified. In addition, the instruction data is based on the identified instruction in the device 120 generated.

Further, one or more embodiments may be described by the following: A computer implemented method for identifying instructions in voice recorded in a vehicle is provided. The method includes providing a list of instruction words and a list of topic words relating to topics of conversation. A unit of speech is recorded in a vehicle and compared with the instruction words and the topic words in the vehicle. When a topic word is identified, the topic word is added to a current topic list. When an instruction word is identified, information relating to the recorded speech unit and the current topic list is transmitted to a remote server. In the remote server, an instruction is identified based on the transmitted information and instruction data is generated based on the identified instruction. The instruction data are transmitted from the server to the vehicle and processed in the vehicle.

As one skilled in the art will recognize, aspects of the present invention can be implemented as a system, apparatus, method, or computer program product. Accordingly, aspects of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, memory-resident software, microcode, etc.), or an embodiment combining software and hardware aspects, all of which are generally referred to herein as a " Unit ”, a“ module ”, a“ system ”, a“ device ”or an“ element ”.

Functions or steps described in this document can be implemented as an algorithm executed by a microprocessor of a computer. Furthermore, aspects of the present invention may take the form of a computer program product which may be written in one or more computer readable media with, e.g. B. stored, computer-readable program code is implemented.

It is to be understood that the above description is included to illustrate the operation of the preferred embodiments, and is not intended to limit the scope of the invention. From the above description, many variations will become apparent to those skilled in the art which are to be included by the scope of the present invention.

Claims (12)

A computer implemented method (300) for identifying instructions in speech recorded in a vehicle (110), comprising the steps of: - providing (310) a list (220) with instruction words; - providing (320) a list (230) with topic words relating to topics of conversation; - recording (330) a speech unit in a vehicle; - comparing (340) the recorded speech unit with the instruction words and the subject words in the vehicle; - if a topic word is identified using the comparison in the recorded speech unit, adding (350) the topic word to a current topic list (270); - if an instruction word is identified using the comparison in the speech unit, transmitting (360) information relating to the recorded speech unit and the current topic list from the vehicle (110) to a remote server (130); - identifying (370) an instruction based on the information in the remote server (130); - generating (380) instruction data based on the identified instruction in the remote server (130) and transmitting the instruction data from the server to the vehicle (110); and - processing (390) the instruction data in the vehicle (110). Procedure according to Claim 1 wherein the current topic list (270) includes topics of conversation, the topic words being assigned to the topics of conversation. Procedure according to Claim 2 where the information related to the current topic list is information related to the topic of conversation with most of the assigned topic words. Method according to one of the Claims 1 - 3 , each added topic word being assigned a timestamp indicating the time the topic word was added to the current topic list. Procedure according to Claim 4 wherein an added topic word is removed from the current topic list (270) when a time difference between a current time and its associated timestamp is longer than a predefined time period. Procedure according to Claim 4 wherein all added topic words are removed from the current topic list (270) when a time difference between a current time and a time stamp of an added keyword that was last added is longer than a predefined length of time. Method according to one of the Claims 1 - 6th wherein the list (230) of subject words is generated in the vehicle (110). Computer program, wherein the computer program is designed to perform the method (300) according to one of the preceding claims perform while the computer program is running on a computer. Computer-readable storage medium which stores a set of code executable by a control unit which, when executed in an apparatus on a processing system or in a digital signal processor, comprises the method steps of any of Claims 1 - 7th performs. Apparatus (120) for identifying instructions in speech recorded in a vehicle (110), comprising: - a storage medium (210) with a list (220) with instruction words, a list (230) with topic words relating to topics of conversation and a current topic list (270); - a voice recording device (240) which is arranged to record a unit of speech; - A transceiver (250) which is set up to exchange information between the device (120) and a remote server (130); and - a processor (260) which is set up to: (i) comparing (340) the recorded unit of speech with the instruction words and the topic words; (ii) if a topic word is identified using the comparison in the speech unit, adding (350) the topic word to the current topic list (270); (iii) if an instruction word is identified using the comparison in the speech unit, instructing the transceiver (250) to transmit (360) information regarding the recorded speech unit and the current topic list (270) to the remote server; (iv) processing (390) instruction data received by the transceiver (250) from the remote server (130) and generated in the remote server (130) based on the information. Vehicle (110) with a device (120) according to Claim 10 . A system (100) for identifying instructions in speech recorded in a vehicle (110), comprising: - a device (120) after Claim 10 or a vehicle (110) Claim 11 ; and - a remote server (130) arranged to: (i) receive information relating to the recorded speech unit and the current topic list (270); (ii) identifying an instruction based on the information in the remote server (130); (iv) generating instruction data based on the identified instruction; and (v) transmitting the instruction data to the vehicle (110).

Images