FR2835087A1 - CUSTOMIZING THE SOUND PRESENTATION OF SYNTHESIZED MESSAGES IN A TERMINAL - Google Patents

Abstract

To customize the sound presentation of messages synthesized in a terminal (1), acoustic characteristics (CV) describing a voice (V) are selected from a catalog of acoustic characteristics pre-memorized in a server (2) in order to transmit them to a device. speech synthesis (3). A textual message (MT) which can be selected in the terminal is synthesized in the equipment depending on the selected acoustic characteristics into a voice message (MS) which is transmitted to the terminal for listening. At least one sound effect (B) can be selected in the server to be mixed with the voice message.

Description

that the serrated screw B acts on the rope by means of a lever C.

  The present invention relates to the audio presentation of messages in a terminal. The messages are initially textual, then synthesized by voice by a means of internal voice synthesis or

external to the terminal.

  o Currently, interactive voice servers or any other means of voice synthesis accessible via a server broadcast messages that result from the voice synthesis of text messages on the basis of artificial or natural voice models. With the exception of pre-filtering and adjustment of the bass and treble in the audio means of most terminals, such as television receivers, radio receivers or personal terminals of the computer type or digital assistant or telephone or radiotelephone terminal, the users of these terminals consult the voice servers all listen to the same voices for the dissemination of synchetised wise messages without any influence

personal on them.

  Because of messages broadcast with a single voice by a given voice server, messages are not

  not always very well perceived by some users.

  The present invention aims to remedy the drawbacks of “single-voice” servers of the prior art so that each user personalizes the acoustic context of the voice messages broadcast by the servers, and thus makes them more intelligible and perceptible, and therefore more

  familiar, broadcast voice messages.

  To this end, a method for personalizing the sound presentation of messages synthesized in a terminal, is characterized in that it comprises steps of selecting acoustic characteristics describing a voice from a first catalog of acoustic characteristics prememorized in a server means in order to transmitting them to a voice synthesis means, and synchronizing a text message in the voice synthesis means depending on the acoustic characteristics selected in a voice message which is transmitted to the terminal. Preferably, the invention completes the selection of the acoustic characteristics to compose a voice by a step of selecting the acoustic characteristics describing a sound effect in a second catalog of acoustic characteristics prememorized in the server means in order to transmit them with the selected acoustic characteristics describing the voice by means of text-to-speech so that the latter transmits the voice message

  mixed with the sound effects selected at the terminal.

  Instead of selecting acoustic characteristics to describe and thus compose a voice and possibly a sound effect, the terminal user can directly select a voice in the first catalog and possibly a sound effect in the second catalog, or even select a combination which is stored. in a third catalog at least in the terminal and which includes at least one voice and possibly at least one sound effect in order to synthesize any text message depending on the acoustic characteristics of the

combination.

  The selection of voices and possibly sound effects is preferably accompanied by a selection of characteristics of a visual presentation, which can be a wallpaper and / or a facial animation, in a fourth catalog prememori ed in the server medium. in order to transmit it to the terminal and display the visual presentation in the terminal in synchronism with the

  reproduction of the voice message in the terminal.

  The invention also relates to a system for personalizing the audio presentation of messages in a terminal, for implementing the method of the invention. The system is characterized in that it comprises a server means for storing acoustic characteristics describing voices, a voice synchesis means in which voices are described depending on acoustic characteristics, and an application means in the terminal for selecting from the server means of the acoustic characteristics describing a voice so that the voice synthesis means synthesizes at least one text message according to the selected acoustic characteristics into a voice message transmitted to the terminal. Preferably, the server means also stores acoustic noise characteristics to be selectively mixed with the voice described. Other characteristics and advantages of the present invention will appear more clearly on

  reading the following description of several

  preferred embodiments of the invention with reference to the corresponding appended drawings in which: - Figure 1 is a schematic block diagram of a system for the audio presentation of messages according to a preferred embodiment of the invention; and FIG. 2 is an algorithm of a method for the audio presentation of messages according to the invention. With reference to FIG. 1, a sound presentation system for messages according to the invention essentially comprises a user terminal 1 provided with at least one loudspeaker or one earpiece, a central sound server 2 and a text-to-speech equipment 3. The system is based on an architecture of the client-server type between the terminal 1 associated with the equipment 3 and the central server of

sounds 2.

  For example, the user terminal 1 is a personal computer or a personal digital assistant, or else a television or intelligent radio receiver, or a landline telephone or a mobile cellular radio telephone. The voice synchesis equipment 3 is connected to the terminal 1 by a conventional link 4 of the wired or proximity radio type. As a variant, the equipment 3 is removably integrated, like a card, in the terminal 1. On the other hand, the terminal 1 is connected to the central server 2 via an access network 5 corresponding to the type of terminal and a packet network 6 such as

the internet network.

  The voice synthesis equipment 3 essentially comprises a buffer memory 30 for storing a text message MT to be synthesized, and preferably at least one test text TE to be synthesized, an analyzer 31 of the phonetics and of the prosody of the text to be synthesized. synChetize, a voice synthesizer 32 proper, and a generator 33 generating an acoustic model according to acoustic characteristics CA delivered by the terminal 1 and supplied by the central server 2. The functional elements 30 to 33 schematically represent the voice synthesis equipment for a better understanding of the invention and may correspond

to software modules.

  The invention relates more particularly to the third module 33 which defines an acoustic model in dependence on the parameters, in particular such as values, CA characteristics of a sound, such as a voice possibly mixed with a sound effect, in order to apply to this model of the predetermined rules for synchronizing a textual message MT transcribed phonetically and prosadically in the analyzer 31 into a synchronized voice message MS transmitted by the synchronizer 32 to the terminal 1. According to another variant, the characteristics CA received in the generator o allow to select acoustic units which are concatenated according to predetermined rules in the synthesizer 32 in order to reproduce vocally an analyzed message

  phonetically and prosodically in the analyzer 31.

  Whatever the type of speech synthesis implemented in the synthesizer 32, the speech synthesis is defined in dependence on acoustic characteristics CA processed in the module 33 and selected by the terminal 1. As will be seen

  in the following description with reference to

  FIG. 2, the terminal 1 supports an application for personal sound presentation of sound 1 0 to select the acoustic characteristics AC of a sound to be composed or to select a fingerprint of a sound described by acoustic characteristics

in the central server 2.

  Text messages MT temporarily stored in memory 10 to synthesize them are provided by the user of terminal 1, either by entering them with the keyboard or by voice recognition in the terminal, or by reading them in the memory of terminal s '' They were pre-registered in the terminal, or even where they are downloaded from servers of servers across networks 5 and 6. Preferably, the test message TE contained in memory 30 is pre-selected by the user of terminal 1 and therefore known by the user to test a user-defined and modalized voice in the equipment of

vocal system 3.

  According to another variant of the architecture of the system, at least one voice synchronization device 3 is integrated in the central server 2 and shared by several user terminals 1 for which user identifiers IDU are associated respectively with test messages MI respectively. The central server 2 is then analogous to an interactive voice server in which 2s of the synthesized voices and their characteristics can be selected by users to listen to voice or multimedia messages. In this variant, the exchanges, in particular of commands of acoustic character and of voice message broadcasting, are carried out through the access network 5 and the packet network 6 between the terminal 1 and the server 2, and not also between the server 2 and the equipment 3 via the terminal 1 according to FIG. 1. According to another variant, the central sound server 2 is distributed into several central servers in each of which one or more catalogs of

  files defined below can be viewed.

  s The central server 2 essentially comprises three catalogs of sound files V (CV, AV), B (CB, AB) and C (CC, AC) from which a terminal user can draw to personalize the sound presentation. of voice messages reproduced in his terminal. All these files can be selected from terminal 1 using

the application 10.

  The first catalog of files V (CV, AV) relates to voices whose vocal imprints were recorded and analyzed in order to memorize the essential acoustic characteristics CV of these voices. For example, the acoustic characteristics describing a predetermined voice V and contained in a file of the first catalog relate to the male or female sex, age in the form of a period relating to childhood or adolescence or adulthood or old age, prosodic characters such as successive durations of 2s syllabic segments, the emphasis in particular on the accent on phrase components, the laryngeal and fundamental frequencies relating to the pitch of the voice (in English "pitch" ), the deUit or the rhythm of speech which can be slow or fast or intermediate, the level of sound expressed in decibels, etc. The file of a predetermined voice V also contains AV attributes specific to each voice which are optional and which concern the owner of the voice such as a user or a company or an organization as a collective user, and / or access restrictions to the voice file so that it can be distributed and used by predetermined users who are identified by IDU identifiers entered in a list of UV users authorized to use the predetermined voice, or characteristics defining a profile of user that a user must present to access the user of the voice, and / or a remuneration o of the use of the voice which can be possibly free, and any other characteristic contributing to

  market the predetermined voice.

  The second catalog of files B (CB, AB) relates to sound effects B which can be sound effects, special sounds or musical pieces of which one or more can be selected by the user of the terminal 1 in order to be superimposed to the voice selects with which a text message is synthesized. Each sound effect B is defined, like the voices in the first catalog, by acoustic characteristics CB and, where appropriate, is associated with attributes AB and a list of users.

authorized UB.

  For practical reasons of downloading, the different sound effects in the second catalog are preferably made up of small, chained and looped sound files. One or more linked sound files can be downloaded

in the terminal.

  The third catalog of files C (CC, AC) relates to files of combinations of sounds which each result from acoustic characteristics CC combining the acoustic characteristics CV and CB at least of a voice V and possibly at least of a sound effect B . or several combinations of voices and

  possibly sound effects distributed in time.

  Each combination is thus defined by CC acoustic characteristics and associated with

  attributes AC and to a list of authorized UC users.

  s A terminal user can thus define an audio program which is divided into various periods during which combinations of voice and sound effects will respectively personalize portions of a text message MT to be synchronized. The attributes 10 of a combination notably define the lengths of period s for respective combinations of sound sounds, as well as the start time of these periods relative to the start time of a message. i5 Preferably, the central sound server 2 comprises a fourth PV catalog (CPV, APV) relating to visual presentations PV of text messages to be synthesized each defined by CPV characteristics and APV attributes. The CPV characteristics of a visual presentation relate to a wallpaper, or more or less animated images, or more particularly the face of a head of an animator whose eyes and mouth at least are animated according to the pronunciation of the synchronized voice message by means of a facial animation engine implemented in the user terminal. The whole animator's head or elements such as eyes and mouths can be chosen from the fourth catalog. As in previous catalogs, APV visual presentation attributes define the owner of the visual presentation on the PV, access restrictions for example associated with a list of UPV authorized users, or remuneration

  relating to the selected visual presentation.

  As a variant, the four catalogs defined above

  above are distributed in respective servers instead of being centralized in a single server 2, and / or are declined for example by geographical region in order to offer in particular voices and sound effects adapted to regional customs or

  and reduce response time.

  Reference is now made to FIG. 2 for describing the main steps E1 to E14 of the method for personalizing the audio presentation of synchronized messages MS from the user terminal 1. The terminal 1 is designated by an identifier IDU which may include a number phone or an Internet Protocol (i5) IP address if applicable

a confidential access code.

  Depending on the type of terminal, the commands relating to selections made during the course of the process correspond to the pressing of a key on the terminal keyboard, for example translated into a DTMF (Dual Tone MultiFrequency) code for a telephone. or a radio recorder or a command specific to a communication protocol or user graphical interface, or else correspond to a voice command recognized by a voice recognition means included in the terminal 1 and / or the server 2. The various selections are preferably assisted by pages displayed in the terminal, when the latter has a display or a display screen, the dialogue between the terminal 1 and the server 2 thus being carried out in a known manner. As a variant, the dialogue between the terminal 1 and the central server 2 is carried out by means of an interactive voice server. The presentation of the catalogs consulted in the server 2 by the terminal 1 is tree-like, that is to say made by means of successive menus and submenus with a return to a main menu. Depending on the personalization application implemented in terminal 1, the user selects the acoustic characteristics of at least one voice V and / or one B sound effect and / or one sound combination C so it di directly in the 0 central server 2, either after download in terminal 1 of a part of the catalogs relating to the files made available to the public, or whose

  access is authorized for this user.

  FIG. 2 illustrates a prAfArA example of steps composing the personalization procAdA according to

  the invention, both for the above variants.

  In a conventional manner at Step E1, the user in front of the terminal 1 opens a session of the application 10 relating to the sound presentation of synthesized messages so that the terminal 1 calls the central server 2 whose address TP is stored in the terminal. If the application allows it, the user selects in a menu the catalogs or categories of sound in the catalogs and the 2s downloaded from the server 2 in the terminal 1 in order to make various selections according to the following steps to build a sound presentation customizes. Otherwise, the following selection steps are carried out through a question-and-answer dialogue between terminal 1 and server 2, which is built as the user selects a specific sound presentation. We are

  Refer to this second variant below.

  In the next step E3, the application 10 asks the user if he wishes to select at least one of his favorite combinations determined and stored previously, possibly in association with the identifier IDU in the server.

  2, if they exist in the third catalog.

  Otherwise, sound characteristics are selected below by the user in order to constitute a combination of sounds personalizing the vocal presentation of MT text messages to be synchronized by

  text-to-speech equipment 3.

  o In the next step E4, the user of the terminal 1 selects acoustic characteristics CV by validating the parameters thereof so as to constitute a personalized voice V in the first catalog of voice files. As a variant, instead of selecting acoustic characteristics of voice V, the user selects a voice V from among several authorized voices in the first catalog, each of which is designated by a name and

  a brief description of the features

  acoustics of it. In step E4, the application optionally offers the user to further personalize the sound presentation of messages by recording a predetermined voice print, for example a predetermined sentence pronounced by the user. If the user wishes to mix the voice defined by the selected characteristics CV or the voice previously selected with one or more sound effects B. the user selects in step E5, in a manner analogous to step E4, acoustic characteristics CB in the second catalog of sound effects files so as to determine one or more sound effects, or directly selects one or more authorized sound effects defined each by predetermined acoustic characteristics. After steps E4 and E5, the voice and possibly noise characteristics selected directly or indirectly are transmitted to the acoustic model generator 33. Then a test text TE is optionally selected in step E6 so that the selected test text is used test for voice synthesis in the synthesizer 32, depending on an acoustic model defined by the selected characteristics CV in the generator 33, before definitively validating the choice of acoustic characteristics of the selected combination of voice and sound effects CS = CV + CB selects in the previous steps E4 and E5. The TE test text selected: 5 in step E6 can be a digitized text or combination of texts, prerecorded in the terminal 1 or the buffer memory 30 of the equipment 3, or entered directly by the user in the terminal 1, or downloaded from one or more servers of at least textual items via the terminal 1 in the memory 30 of the equipment 3. The test text TE is preferably stored in the memory 30 of the equipment 3 in particular for steps of test in

  subsequent audio presentations.

  In the next step E7, the test text TE read in the memory 30 is analyzed by the analyzer 31 and synthesized in the synthesizer 32 as a function in particular of the acoustic model defined by the acoustic characteristics of voice CV selected in step E4 with possible mixing of sound effects B selected in step E5. The voice message MS produced by the synthesizer 32 is transmitted to the

  terminal 1 so that the user can listen to it.

  If the user is not satisfied with the acoustic characteristics of the voice message produced, the application 10 suggests to him in step E8 to modify, that is to say to add or remove or correct an acoustic characteristic CV of the voice described V or CB of the sound or sounds described B selected in steps E4 and E5, returning to step E4. Finally, after possibly one or more repetitions of steps E4 to E8, the terminal 1 and the server 2 store the acoustic characteristics [CV + CB] of the selected combination 0 CS in step E9, preferably by associating it with

the user identifier IDU.

  Optionally, in step E10, the application 10 offers the user a selection of CPV characteristics in the fourth catalog of visual presentation files in order to select in step E101 a visual presentation such as a background.

  screen and / or facial animation.

  In the next step E11, the terminal 1 and the server 2 store the visual presentation characteristics CPV possibly selected in step E101 in association with the combination of acoustic characteristics of voice and sound effects [CV + CB] selected in steps E4 to E8. Then at the following steps, in particular E121, E122 and E123, the application 10 invites the user of the terminal 1 to select one or more parameters, in particular time and / or documentary parameters, personalizing the use of the combination of sounds CS composed in the preceding steps. In step E121, the user indicates two dates of

  broadcast about the selected combination CS.

  In practice, the user indicates a start date of broadcast and / or an end date of broadcast of the selected combination. One or more broadcast periods can thus be selected to make the selected combination accessible during these periods. By associating such periods with various selected combinations, audio programs are created. The audiovisual programs created are preferably displayed in the terminal with their respective time positions. All the preceding or following temporal data o are expressed in years,

  month, day, hour, minute and second.

  In step E122, the application 10 proposes to the user to determine the instant of start of introduction of the selected combination CS as well as the duration of the latter in relation to the instant of start of listening d a voice message MS synthesized in the equipment 3 in order to constitute with other selected combinations an audio program. Optionally, the start time and the duration of diffusion of the selected combination CS are chosen randomly by the application 10. As a variant, several combinations are selected to constitute a series of combinations which is

repeated periodically.

  In step E123, the application 10 offers the user to associate predetermined documents with the selected combination CS. Each of these items is identified by an identifier which can be a name and / or an address such as a URL (Uniform

  Resource Locator) read from a website server.

  The association of a document with one or more combinations can be made available to any user. After steps E121 and / or E122 and / or E123, or after step E3 having selected a so-called "favorite" combination already stored in the server 2 or the terminal 1 and accessible to the user, the application invites the user of the terminal 1 to listen to a text message MT of his choice in order to transmit it to the text-to-speech equipment 3 which has received the acoustic characteristics CV and CB of the selected or preferred combination. The voice message MS resulting from the voice synthesis of the selected text message MT is listened to by the user o simultaneously with a possible visual presentation such as a facial animation whose CPV characteristics were selected in step E10 and which is displayed in the terminal. Then if the user wishes to modify i 5 acous tic and / or visual characteristics of the presentation of the synthesized message, as indicated in step E14, he again selects acoustic voice characteristics. and / or sound effects and possibly visual presentation characteristics from step E4. Otherwise, the session of the application 10 at least between the terminal 1 and

  the equipment 3 is completed in step E15.

Claims (12)

  1 - Method for personalizing the sound presentation of messages synthesized in a terminal (1), characterized in that it comprises steps of selecting (E4) acoustic characteristics (CV) describing a voice (V) in a first catalog of acoustic characteristics memorized in a server means (2) in order to transmit them to a voice synthesis means 0 (3), and to synthesize (E7) a text message (MT) in the voice synthesis means depending on the selected acoustic characteristics into a voice message (MS) which is transmitted to the terminal (1).   2 - Method according to claim 1, comprising a step (E5) of selecting acoustic characteristics (CB) describing a sound effect (B) in a second catalog of acoustic characteristics prememorized in the server means (2) in order to transmit them with the selected acoustic characteristics (CV) describing the voice (V) by means of voice synthesis (3) so that it transmits the voice message (MS)   mixed with the sound effects selected at the terminal (1).   3 - Method according to claim 1 or 2, according to which the step of selecting (E4, E5) acoustic characteristics describing a voice (V) or a sound effect (B) is replaced by a step of directly selecting a voice in the first   catalog or sound effects in the second catalog.   4 - Process in accordance with in accordance with one   any of claims 1 to 3, comprising a   step (E6) of selecting the text message (TE) to synthesize. Method according to Claim 4, in which the text message to be synthesized is transmitted by the terminal (1) and stored in the means for speech synthesis (3).   6 - Method according to claim 4, according to which the text message (TE) to be synthesized is selected in a text document server in order to download it via the terminal (1) in the voice synthesis means (3).   7 - Process according to any one of   claims 1 to 6, comprising after listening   (E7) of the voice message (MS) transmitted by the voice synthesis means (3) to the terminal (1), a step (E8) of adding or removing or correcting an acoustic characteristic (CV, CB) for describe to. molns the VOlX.   8 - Process according to any one of   claims 1 to 7, comprising a step (E3) of   select at least one combination (C) stored in a third catalog at least in the terminal and comprising at least one voice (V) and possibly at least one sound effect (B) in order to synthesize any text message (MT) in dependence   acoustic characteristics of the suit.   9 - Process in accordance with any of the   claims 1 to 8, comprising a selection (E10,   E101) of characteristics of a visual presentation (CPV) in a fourth catalog stored in the server means (2) in order to transmit them to the terminal (1) and to display the visual presentation in the terminal in synchronism with the reproduction of the   voice message (MS) in the terminal.   - Process according to any one of   claims 1 to 9, further comprising a step   (E121, E122, E123) to select at least one of the following parameters customizing the use of at least the voice (V) declute by the selected acoustic characteristics: date and period of broadcast of the selected voice, instant of introduction and duration of the selected voice relative to the instant of the start of a voice message i5 synchronized in the voice synthesis means (3), identifier of documents to be associated with the voice, combination of sounds, including the selected voice, or a series of combinations.   11 - Process according to any one of   claims 1 to 10, comprising beforehand in   the server means (2) a definition of attributes specific at least to voices and relating to the property and / or a restriction of access and / or a   remuneration for the use of votes.   12 - System for personalizing the sound presentation of synDhetized messages in a terminal (1), for implementing the method according to one   any of claims 1 to 11, characterized in   what it includes server means (2) for memorizing acoustic characteristics (CV) describing voices (V), voice synchesis means (3) in which voices are described depending on acoustic characteristics, and application means (10) in the terminal (1) for selecting in the server means (2) acoustic characteristics describing a voice so that the voice synthesis means synthesizes a text message (MT) according to the selected acoustic characteristics in one   voice message (MS) transmitted to the terminal.   13 - System according to claim 12, wherein the server means (2) also stores o acoustic characteristics (CB) of sound effects (B)   to be mixed selectively with the voice described.   14 - System according to claim 12 or 13, wherein the voice synchesis means is an equipment (3) located near or integrated in the terminal (1).   - System according to claim 12 or 13, wherein the voice synchesis means (3) is