EP1344211B1 - Device and method for differentiated speech output - Google Patents

Abstract

The invention relates to a device and to a method for differentiated speech output. The systems available in a motor vehicle, such as on-board computer, navigation system and others can be linked with a speech output device. The speech output of different systems can be differentiated by voice characteristics.

Description

The present invention relates to a differential speech output apparatus and method, systems for use with the speech output apparatus, and combinations of a speech output apparatus having at least two systems, particularly for use in a vehicle.

In vehicles, individual systems are used which have an acoustic man-machine interface for voice output. In these systems, a voice output module is assigned directly in each case. The speech-producing methods used are usually based on pulse-code modulation (= PCM), with subsequent compression (eg MPEG) being able to be connected. Other systems use speech synthesis methods, which form words and sentences mainly through the compilation of syllable segments (phonemes) (signal manipulation).

In the mentioned speech output methods, there is also a speaker dependency, which requires, as the word or text extent is expanded, to repeatedly strive for the same human speaker for recordings. Furthermore, PCM methods, as well as high-quality phoneme synthesis by signal manipulation, require considerable memory space to store texts or syllable segments. In both methods, the memory capacity increases significantly when different national languages are to be output.

Furthermore, methods are known which are based on a full synthesis of the language. In particular, methods are known which implement the human vocal tract as electrical equivalent and work with a tone generator and several downstream filters (source filter model). A device operating according to this method is a so-called formant synthesizer (eg KLATTALK). Such a formant synthesizer has the advantage that the tuning characteristic properties can be influenced.

The EP-A-0 901 000 describes a device for processing messages with receiving means for receiving sent messages, a memory for storing a plurality of different articulations (tone of voice respectively) and assignment means for assigning an articulation of the plurality of articulations to at least one received message. Another articulation is assigned to another received message and output means issues the first message with a first articulation and the second message with a second articulation.

RUTLEDGE JC ET AL: "SYNTHESIZING STYLED SPEECH USING THE CLAD SYNTHESIZER" PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON ACOUSTICS; SPEECH; AND SIGNAL PROCESSING (ICASSP): DETROIT; MAY 9-12, 1995. SPEECH, NEW YORK, IEEE, US, Vol. 1, 9 May 1995 (1995-05-09), pages 648-651, XP000658077 ISBN: 0-7803-2432-3 He is generally concerned with the working and experimental experience in building vocal tract-based vocal synthesizers (Klatt Synthesizers) for the synthesis of different speech styles with different vocal characteristics.

The invention is in particular the object of providing a central speech output device with a plurality of systems, in which a single speech generator with a small parameter memory is driven by the systems.

This object is achieved by the device with the features of claim 1. Advantageous embodiments are the subject of the dependent claims.

The invention has the advantage that speech outputs for different systems are possible with a single speech output device or speech synthesis device, wherein each system can be identified by vocal characteristic differences.

According to a preferred embodiment of the invention, a parameter set is assigned for each system, which is used by the speech synthesis device in a speech output from this system. For example, a first parameter set for an on-board computer, a second parameter set for a navigation system, a third parameter set for traffic information, a fourth parameter set for a TTS system (text to speech system), such as e-mail and one or more additional parameter sets for additional Systems provided.

Depending on the assigned parameter set, the speech synthesizer generates the speech output, for example, with a soft female voice, e.g. for voice output of a navigation system, or with a hard male bass voice, z. B. for the voice output of traffic reports.

According to a preferred embodiment of the invention, a method and apparatus is used for full speech synthesis, preferably a formant synthesizer. The control parameters for the synthesizer are divided into classes. A class of dynamic parameters controls the articulation, as the movement of the language tract in speaking. A second class of static parameters controls speaker characteristic features, such as the generator fundamental frequency and fixed formants, in a child, a woman, or a woman a male speaker formed by the different geometric dimension of the language tract.

In an extended model of the formant synthesizer, a separate generation of voiced and unvoiced sounds is possible. In this case, additional parameters or additional resonators or attenuators can be switched on or the dynamic parameters for the articulation can be influenced.

The device according to the invention or the method according to the invention can be used in particular in systems of a vehicle. Each system has two ways to control the voice output for a voice output. The first possibility of the speech output comprises sending an output of control commands for the language articulation, wherein the sequence of control parameters for words, sentences and sentence sequences are stored in the system. The second way to control the speech output is via a second output, which switches a parameter set, which determines the speaker characteristic.

Alternatively or additionally, it is also possible to store this parameter data record directly in the system and to load the parameter data record into the speech synthesis device when a voice output is required.

According to a further preferred embodiment, which can be used alternatively or in addition to the above embodiments, in order to distinguish the sources of information, i. H. of the systems that perform a speech output, the generator and formant parameters are additionally dynamically changed. This allows audible differences in prosody to be achieved, such as the duration and / or stress on syllable segments and / or sentence melody. In particular, a prosodic modulation depending on z. B. be used by a traffic situation or a traffic situation for the voice output of announcements. Finally, the explosiveness of information can be expressed by modulating the voice.

The invention has the advantage that z. B. in a vehicle only a single voice generator with a small parameter memory of several information sources driven can be. The sources of information can be equipped with different voice characteristics.

When using a Vollsyntheseeinrichtung, z. As a vocal tract synthesis device shows that the method is speaker independent and no high-quality studio records are needed.

In the case of an extended formant synthesizer, emotional expression in the voice can also be given according to the invention.

Predefined parameter templates make it easy to change the tuning characteristics. The method is also suitable for the implementation of free texts in speech (text to speech), z. B. reading aloud e-mail.

The invention is explained below with reference to an embodiment and the drawing.

Fig. 1 shows a schematic diagram of a preferred embodiment of the invention for differentiated speech output with multiple systems according to the invention.

In the Fig. 1 illustrated preferred embodiment of the invention comprises a speech output unit 1 with a speech synthesis device 10, which in the example is a vocal tract synthesis module based on a full synthesis of the language. For example, a formant synthesizer such as KLATTALK can be used. The speech synthesizer 10 is connected to an amplifier 12 whose output 14 provides an audio signal that outputs speech through a speaker (not shown). The speech synthesis device 10 is assigned to N parameter sets 21, 22 to 2N, which in the example shown are stored in a memory 20 of the speech output unit 1. Furthermore, N systems 31, 32 to 3N are shown, which are each connected to the voice output unit 1 via a data connection, such as individual lines, a bus system or data channels. Each system can voice over the voice output unit. In detail, an on-board computer 31 with an associated parameter set for the on-board computer 21, a navigation system 32 with an associated parameter set for the navigation 22, a traffic information system 33 with an associated parameter set for the traffic information 23, an e-mail system such as TTS system 34 with an associated parameter set for e-mail 24. Additional systems 3N may be provided with a respective assigned parameter set 2N. In the example shown, it is possible, using a single voice output unit 1, to have the navigation system 32 speak, for example, a soft female voice determined by the parameter set for the navigation system 22. Further, for example, for traffic reports, a parameter set 23 may be provided, with which a hard male bass voice is used in the speech output.

The order of the voice outputs may be sequential in time according to the receipt of the voice output request from the systems. Preferably, higher priority information, e.g. Traffic information in case of danger such as wrong-way driver first issued by voice output. Particular preference is given to information of the highest priority, e.g. Information issued by the on-board computer about malfunction of the vehicle or incipient road smoothness immediately, with a current voice output can be interrupted. The interrupted speech output can then be completed or repeated.

The invention has the advantage that systems with an acoustic display provide the driver without distracting him from his task, as is the case with visual displays to provide information from various systems. The use of a speech synthesis device, which can be used by various on-board computers, can save costs. Compared to previously used speech-producing method in, for example, navigation systems, the storage space requirement can be reduced.

The invention is particularly advantageously used in motor vehicles.

Claims (10)

1. A device for differentiated speech output (1), which can be connected to a first system (31) and at least one further system (32, 33 to 3N), a first voice characteristic being assigned to the speech output of the first system (31) and a further voice characteristic, which audibly differs from the first voice characteristic, being assigned to the further speech output of the further system (32, 33 to 3N), characterised by a speech synthesis mechanism (10), which receives control parameters, which have a first class of dynamic parameters and a second class of static parameters, the dynamic parameters controlling the articulation in accordance with the movement of a speech tract, and the static parameters controlling the voice-characteristic features, the static parameters for the systems being stored as assigned parameter sets in a memory (20) of the speech output device and an assigned parameter set being used by the speech synthesis mechanism (10) for the speech output depending on a selection signal of a system, and the dynamic parameters being stored in accordance with the sequence of words, sentences and sentence sequences in each system.
2. A device according to claim 1, wherein the static parameters have a fundamental generator frequency and/or fixed formants, which preferably correspond to the different geometric dimension of the speech tract in a child, a woman or a male speaker.
3. A device according to claim 2, wherein the generator and/or formant parameters can be changed for the speech output of various systems and audible differences are preferably brought about in the prosody and the duration and/or emphasis of syllable segments and/or the sentence melody.
4. A device according to any one of claims 1 to 3, wherein the speech synthesis mechanism (10) is a formant synthesiser, with which the voice-characteristic properties can be influenced.
5. A device according to claim 4, wherein the formant synthesiser is suitable for separately generating voiced and unvoiced sounds, and wherein additional resonators or attenuators can be switched on, especially using further parameters, and/or the dynamic parameters for the articulation can be influenced.
6. A device according to any one of claims 1 to 5, wherein the speech synthesis mechanism (10) is connected to an amplifier (12) and a speech output takes place by means of an audio output (14) of the amplifier (12).
7. A system for use with a device according to any one of claims 1 to 6, with a first output for outputting dynamic parameters and a second output for outputting a selection signal to switch over a parameter set in the speech output device (10).
8. A system for use with a device according to any one of claims 1 to 6, with an output for outputting dynamic parameters and static parameters, preferably as a parameter set, to the speech output device (10).
9. A combination of a device according to any one of claims 1 to 6, with at least a first and a further system, such as an on-board computer (31), a navigation system (32), a traffic information system (33), an e-mail system (34), or an information system (3N), preferably for use in a vehicle.
10. A method for differentiated speech output using a device according to any one of claims 1 to 6.

Images