JPH037996A - Generating device for singing voice synthetic data - Google Patents

Abstract

PURPOSE:To synthesize a singing voice with an accurate interval by correcting the pitch of pitch data extracted from a song voice into the pitch of the closest scale. CONSTITUTION:A pitch data correction part 7 corrects the pitch of the pitch data found from an actual song voice into the closest pitch in a pitch-scale table 7a. Thus, the pitch data extracted from the song voice is corrected to obtain the same effect with a case wherein the pitch data is inputted from a score. Therefore, the singing voice with the accurate interval can be synthe sized.

Description

[Detailed Description of the Invention] C. Industrial Application Field] The present invention is directed to the creation of song speech synthesis data that is input to a speech synthesis circuit that performs rule synthesis of song speech based on speech data consisting of phonological information and prosody information. It is related to the device.

[Prior Art] In general, a speech rule synthesis method is a method that generates speech based on phonetic and linguistic rules by inputting phonological parameters that control phonological information and prosodic parameters that control prosodic information. Although the amount of information to be processed (external input information, internally stored information) is very small, it is a speech synthesis method that can output synthesized speech of a large number of words and sentences.

On the other hand, the text synthesis method is the ultimate speech synthesis method that performs speech synthesis only from character strings. It automatically creates phonological information and prosody information from character strings and performs rule synthesis based on that information. It will be realized. However, although it is currently being actively researched, the quality of synthesized speech is still low.

Here, text synthesis is divided into two parts: automatically creating phonological information and prosody information from character strings, converting them into audio data, storing them on a storage medium or outputting them to a transmission line, and It is conceivable to extract voice data and divide it into two parts, a part that outputs the voice by rule synthesis.In this way, the part that outputs the synthesized voice can be made compact, that is, it can be made smaller, lighter, and lower in cost. (See Japanese Patent Application No. 62-263298).

By the way, if we consider the case of synthesizing singing voices, the ultimate form in this case would be to input musical scores and output singing voices. Comparing this with the text synthesis method, it is easier to generate prosodic information because musical scores contain not only character strings (lyrics) but also other information such as height information. This means that a product with a simple structure and high quality can be obtained.
177314).

[Problem to be solved by the invention] However, as mentioned above, when inputting musical scores,
In addition to the lyrics, it is necessary to input height information, such as the C■ syllable string and the pitch for each Cv syllable.When inputting this musical score data, consider inputting it from a printed version using an optical reading device. Although the technology for reading printed characters is almost established, the technology for reading musical notes has not been established. Therefore, there is a problem in that at least data related to height must be input manually by a human using a keyboard, making the input work troublesome. Also,
When inputting a musical score, although the amount of information is larger than just a string of characters, the height is determined more uniquely than the notes, but the length is only given as the length for each note.
The length of consonants and vowels is not given, so it must be determined by rules, and the strength is also not given as a specific value, so this is also determined by rules. However, with the current level of technology, only low-quality audio can be obtained.

The present invention has been made in view of the above points, and its purpose is to simplify the data input work by inputting musical score data using only character strings, and furthermore, it is possible to input musical score data using character strings only. As the song voice synthesis data input to the synthesis circuit, voice data for natural song voices is obtained,
Moreover, it is an object of the present invention to provide a song voice synthesis data creation device that can obtain voice data that can synthesize higher quality song voices with accurate pitches.

[Means for Solving the Problems] The song speech synthesis data creation device of the present invention provides song speech synthesis data that is input to a speech synthesis circuit that performs rule synthesis of song speech based on speech data consisting of phonological information and prosody information. In the data creation device, a musical score and a singing voice are input, and a musical score information extraction means extracts character string data, which is phonological information, from the input musical score, and height data, which is prosody information, from the input singing voice; A song voice information extraction means for extracting strength data and length data is provided, and a height data correction section is provided for correcting the pitch of the extracted height data to the pitch of the nearest musical scale. The height data, strength data, and length data obtained are output as audio data.

[Function] The present invention is configured as described above, and is used in an apparatus for creating song speech synthesis data that is input to a speech synthesis circuit that performs rule synthesis of song speech based on speech data. Only character string data is extracted, eliminating the need for keyboard input of height data, making data entry easier.In addition, the system extracts metrical information based on the actual singing voice. This allows higher quality singing voices to be output from the speech synthesis circuit, and since the pitch of the height data extracted from the singing voices is corrected based on the pitch of the pitch-scale table, it is possible to output higher quality singing voices from the speech synthesis circuit. It is now possible to synthesize singing voices with a certain pitch.

[Embodiment] FIG. 1 shows an embodiment of the present invention, in which song speech synthesis data input to a speech synthesis circuit χ that performs rule synthesis of song speech based on speech data consisting of phonological information and prosody information is shown. The creation device Y receives a musical score and singing voice as input, and includes a musical score information extraction means for extracting character string data, which is phonetic information, from the input musical score, and height data, which is prosody information, and strength data, which is prosody information, from the input singing voice. In addition, a height data correction unit 7 is provided to correct the pitch of the extracted height data to the nearest bi-sochi scale. , the corrected height data, strength data, and length data are output as audio data. Here, in the embodiment, the musical score information extraction means is formed by a musical score input section 1 consisting of an optical musical score reading device that automatically reads character string data of the musical score, and a phonetic information extraction section 2. includes a singing voice input section 3 such as a microphone, a feature extraction section 4 that extracts features of the input singing voice,
A segmentation processing unit 5 that divides the feature-extracted singing voice into each phoneme, and a prosody information extraction unit 6 that extracts height data, strength data, and length data of each divided phoneme.
and the extracted height data as pitch-scale table 7a.
a height data correction unit 7 that corrects the height data based on the height data correction unit 7;
The character string data, strength data, length data, and corrected height data are subjected to predetermined encoding in the encoding unit 6 so that they are output as voice data input to the voice synthesis circuit X. 6 In the present embodiment, the musical score and singing voice are input, and the voice data (phonological information and prosody information) is output to the voice synthesis circuit X of the regular synthesis method, and the character string is (phoneme sequence) data is automatically extracted by an optical reader. Note that a human may read the musical score, convert it into data, and input it from a keyboard.

On the other hand, height data, strength data, and length data are extracted from the input singing voice, and for this purpose, it is necessary to segment the singing voice into each phoneme. , at this time, segmentation can be easily performed by using information on phoneme sequences input from the musical score (Proceedings of the Acoustical Society of Japan (March 1986 > 2-
4-1 "Automatic creation of phoneme dictionary by top-down phoneme segmentation J)" Based on the segmentation results, height data and strength data of each mora and length data of each phoneme are extracted.

Next, the height data correction section 7 corrects the pitch of the height data obtained from the actual singing voice to the closest pitch in the pitch-scale table 7a as shown in FIG. That is, the pitch is corrected to a scale that is considered to correspond to the point where the pitch was found. By correcting the height data extracted from the singing voice in this way, the same result as when inputting the height data from the musical score can be obtained.

Each data obtained in the above manner is encoded by the encoding unit 6, stored in a storage medium or output to a transmission line, and used as voice data for the voice synthesis circuit X to produce a natural song with accurate pitch. The audio will be synthesized. In this case, only character string data is read from the musical score, and input of height data from a keyboard is not necessary, so data input work can be easily performed. In addition, since the prosody information is extracted based on the actual singing voice, it is possible to output higher quality singing voices from the speech synthesis circuit, and the pitch of the height data extracted from the singing voices can be changed to -Since it is corrected based on the pitch of the scale table, it is possible to synthesize singing voices with accurate pitches.

[Effects of the Invention] The present invention is configured as described above, and in a device for creating song speech synthesis data that is input to a speech synthesis circuit that performs rule synthesis of song speech based on speech data, phonetic information is extracted from a musical score. Since only certain character string data is extracted, eliminating the need for keyboard input of height data, data entry work can be done easily.In addition, prosodic information can be extracted based on the actual singing voice. This allows higher quality singing voices to be output from the speech synthesis circuit, and since the pitch of the height data extracted from the singing voices is corrected based on the pitch of the pitch-scale table, This has the effect of being able to synthesize singing voices with accurate pitches.

[Brief explanation of the drawing]

FIG. 1 is a block circuit diagram showing a schematic configuration of an embodiment of the present invention, and FIG. 2 is an explanatory diagram of the same operation. X is a speech synthesis circuit, Y is a song speech synthesis data creation device, 1
2 is a musical score input section, 2 is a phonetic information extraction section, 3 is a song voice input section, 4 is a feature extraction section, 5 is a segmentation processing section, 6
7 is a prosodic information extraction section, 7 is a height data correction section, 7a is a pitch-scale table, and 8 is an encoding section.

Claims (1)

[Claims] (1) In a device for creating song speech synthesis data that is input to a speech synthesis circuit that performs regular synthesis of song speech based on speech data consisting of phonological information and prosody information, musical score and singing speech are input. Musical score information extraction means for extracting character string data as phonological information from a musical score, and song voice information extraction means for extracting metrical information such as height data, intensity data and length data from input song voice are provided. In addition, a height data correction section is provided to correct the pitch of the extracted height data to the pitch of the closest musical scale, and the character string data, corrected height data, strength data, and length data are output as audio data. A device for creating song speech synthesis data.