KR100585711B1 - Method for audio and voice synthesizing - Google Patents

Abstract

The present invention relates to an audio and speech synthesis method for recording and synthesizing speech while playing various background audio.

The present invention provides a method of synthesizing audio and speech using a VXML document in a VXML server. Playing the file (file1) specified in the source); Receiving and storing a user's voice as a file (file2) designated by the voice storage file element (dest) of the VXML document; And synthesizing the audio file (file1) and the voice file (file2), and storing them in a file (file3) specified by the synthesis file element (synthesis_file) of the VXML document.

Description

METHOD FOR AUDIO AND VOICE SYNTHESIZING}

1 is a block diagram showing the configuration of a general VXML system.

2 is a diagram illustrating the functions and meanings of 47 tags defined in the conventional VXML 1.0.

Figure 3 is an exemplary view showing the configuration of a synthesis element (<synthesizer>) for recording voice while serving background audio according to the present invention.

4 is an exemplary view showing VXML content for recording voice while playing music using an element (<synthesizer>) according to the present invention.

TECHNICAL FIELD The present invention relates to elements for audio and speech synthesis, and more particularly, to an audio and speech synthesis method that allows a VXML server to record and synthesize speech while playing various background audio.

As the speech recognition technology has been developed to a practical level, voice recognition applications including voice portals are emerging as new issues. In the preparation of the IMT-2000 project, voice recognition-related technology has become an important theme. In this atmosphere, the VoiceXML (Voice eXtensible Markup Language) document format has recently been expected to be the most influential way to realize voice services combining speech synthesis and speech recognition.

For reference, VXML is a document format defined using XML grammar, and the so-called VXML document written in accordance with this document format serves as a scenario for specifying how a conversation proceeds in a voice application.

In order to actually use VXML, three components are required: voice platform, VXML content, and VXML interpreter.

1 is a block diagram showing the configuration of a general VXML system, a document server (Document Server) 110 is a software element for transmitting a VXML document in response to a request of the VXML interpreter, it is possible to use a web server used for homepage operation as it is. The web server is responsible for sending the documents requested by the client. It mainly works to transmit HTML related documents, but there are no special restrictions on the format of sent documents, so it can be used to send VXML related documents without modification.

Using a web server has the advantage of easily connecting VXML voice services to Internet content. In particular, by using the CGI function of the web server, voice input information can be reflected in generating a new document. At this time, if the database is linked using ASP or PHP, real-time processing such as stock quote, weather, and sports relaying is possible.

Unfortunately, HTML-related documents, which occupy most of the existing Internet content, are very different in structure and content and cannot be used as they are. In other words, the web server itself can be used as it is, but the content must be rebuilt.

Next, the Implementation Platform 130 is an element of a voice terminal composed of hardware supporting voice input / output and communication functions, and software supporting voice recognition and voice synthesis functions, and according to the instructions of the VXML interpreter. It not only reports the result by voice input / output such as recognition, audio file output, voice input recording, DTMF (telephone button) input, but also can occur during voice service process such as user's non-response, various timeouts, user disconnection, etc. It also detects and notifies various events.

Next, the VXML interpreter context 120 provides additional functions necessary for interpreter execution, such as interpreter initialization, user profile information processing, interpreter monitoring, and user access management, so that the same interpreter module can be easily ported to various execution environments.

The key element of the overall system component is also the VXML interpreter module. The VXML interpreter analyzes the structure of the VXML document using the built-in XML parser, interprets the contents of the document, and executes the control structure according to the contents. It will serve as a command tower to enable VXML voice service by directing various tasks such as processing various events, handling requests from the VXML interpreter context, and switching to a new working document using a document server.

In general, the markup languages used in the voice Internet include VoiceXML 1.0 and VoiceXML 2.0.

In the case of VoiceXML 1.0, voice is synthesized through a <prompt> element, or a voice file is provided to a user by executing a pre-recorded audio file through a <audio> element. For reference, FIG. 2 illustrates functions and meanings of 47 tags defined in VXML 1.0.

In VoiceXML 2.0, voice is synthesized through elements defined in the Speech Synthesis Markup Language Specification and a <prompt> element, or a voice file is provided to a user by executing a pre-recorded audio file through the <audio> element.

However, in the voice Internet service, the elements used in the VoiceXML 1.0 or 2.0 provide only a hard voice or a single audio, so that voice Internet users can record audio by singing along with accompaniment, such as karaoke. There is a disadvantage in that it is not possible to provide various voice services such as synthesizing two audio files by inputting voice while listening.

Accordingly, the present invention provides audio and speech synthesis that enables recording and synthesis of speech while reproducing various background audio using VoiceXML, which is a markup language used for the voice Internet, to improve the conventional problems. The purpose is to provide a method.

In order to achieve the above object, the present invention provides a method for synthesizing audio and speech using a VXML document in a VXML server, and detecting an element indicating a synthesis of audio and speech in the VXML document. Playing the file (file1) specified in the audio playback file element (Source) of the VXML document; Receiving and storing a user's voice as a file (file2) designated by the voice storage file element (dest) of the VXML document; And synthesizing the audio file (file1) and the voice file (file2), and storing it as a file (file3) designated by the synthesis file element (synthesis_file) of the VXML document.

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the present invention will be described.

FIG. 3 is a diagram illustrating a configuration of a synthesis element (<synthesizer>) for recording voice while serving background audio according to the present invention, wherein the <synthesizer> element is a symbol of a <form> element in a DTD (Data Type Definition). Located as a child element, if you look at the DTD for the <synthesizer> element, there are elements belonging to <prompt>, <audio>, and '% event.handler'.

Referring to the attributes representing the characteristics of the <synthesizer> element with reference to FIG. 3, the attributes are composed of 'id', 'type', 'beep', 'maxtime', 'source', 'dest', and 'synthesis_file' attributes. The meaning of each property is explained as follows.

First, 'id' represents the name of the <synthesizer>, and 'type' represents the MIME format of the audio file to be synthesized, such as voice. This value applies equally to the MIME format of the voice file you are recording and the MIME format of the file to be synthesized.

Next, 'beep' generates this 'beep' sound before entering the recording when this value is 'true'. The default value is 'false'.

Next, 'maxtime' refers to the maximum time that can be recorded, 'dest' refers to the Uniform Resource Identifier (URI) of the voice file recorded by the user, and 'source' refers to the URI of the audio file to be synthesized, such as voice. Indicates.

Next, 'synthesis_file' represents a URI of a file to be stored by synthesizing a predetermined audio file and a voice recorded by a user, and 'type', 'source', 'dest', and 'synthesis_file' are essential attributes. .

Then, <prompt> and <audio> coming to the child element of the <synthesizer> element are used to give voice or audio to the user before recording the voice, and '% event.handler' Used to deal with errors that may occur.

In other words, the VXML interpreter 120, when interpreting a VXML document, if there is a <synthesizer> element, records the voice with the name specified in the 'dest' attribute and then synthesizes it with the audio file specified in the 'source' attribute. The request is made in the implementation of the voice platform (Implementation Platform) 130, the voice platform is stored with the name specified in the 'synthesis_file' attribute.

FIG. 4 is an exemplary view showing VXML content recording a voice while playing music using an element (<synthesizer>) according to the present invention. The user hears a sentence "At the tone, please say your greeting", for example. If you hear a beep, you can record your voice. At this time, the user records while listening to the 'sample2.wav' audio specified in the 'source' property.

At this time, the voice to be recorded is stored as a 'sample1.wav' file specified in the 'dest' property, and the VXML server (not shown) synthesizes 'sample1.wav' and 'sample2.wav' to display 'sample3.wav'. The final audio file is created. If the user does not input any voice, the voice output 'I didn't hear anything, please try again.'

As described above, the present invention defines a tag capable of simultaneously performing a kind of music reproduction and voice synthesis, which is not previously defined. The <synthesizer> element allows a user to easily apply an effect such as karaoke by applying a VXML grammar. You can do it.

As described in detail above, the present invention has an effect of not only providing a hard voice or a single audio service to voice internet users, but also recording and synthesizing the voice while playing various background audio.

Claims (5)

In performing the synthesis of audio and speech using a VXML document in a VXML server, Detecting a synthesizer indicative of the synthesis of audio and voice in the VXML document, and playing a file designated by an audio reproduction file element of the VXML document; Receiving and storing a user's voice as a file (file2) designated by the voice storage file element (dest) of the VXML document; And synthesizing the audio file (file1) and the voice file (file2), and storing them as a file (file3) designated by the synthesis file element (synthesis_file) of the VXML document. . The method of claim 1, wherein the VXML server, And generating a beep when a beep element is present in the element of the VXML document for audio synthesis, and recording up to a time specified by a maxtime element. The audio and speech synthesis method of claim 1, wherein each of the elements 'dest', 'source', and 'synthesis_file' is expressed in a 'URI' format. The method of claim 1, wherein the VXML document, Elements (<prompt>, <audio>) for presenting voice or audio to the user; An audio and speech synthesis method comprising an element (% event.handler) for error handling that may occur during voice recording. The method of claim 1, wherein the VXML document, An id element representing a name of a synthesizer to synthesize the audio and voice; And a type element indicating a mime format of an audio file to be synthesized, such as the voice.

Images