KR20220071523A - A method and a TTS system for segmenting a sequence of characters - Google Patents

Abstract

An objective of the present invention is to provide an artificial intelligence-based speech synthesis technology capable of realizing a natural speech as if a real speaker is speaking. According to an embodiment of the present invention, provided is a method for segmenting a sequence of characters composed of a specific natural language, which comprises: a step of generating a first group including a plurality of subsequences by segmenting the sequence based on at least one punctuation mark included in the sequence; a step of generating a third subsequence by merging the first subsequence and a second subsequence adjacent to the first subsequence in a case in which the first subsequence included in the first group is shorter than a first threshold length; generating a second group by updating the first group based on the third subsequence; generating a plurality of fifth subsequences by segmenting a fourth subsequence according to a predetermined criterion in a case in which the fourth subsequence included in the second group is longer than a second threshold length; and generating a third group by updating the second group based on the plurality of fifth subsequences.

Description

{A method and a TTS system for segmenting a sequence of characters}

A method for segmenting a sequence of characters and a speech synthesis system are provided.

Recently, with the development of artificial intelligence technology, interfaces using voice signals are becoming common. Accordingly, research on a speech synthesis technology capable of uttering a synthesized voice according to a given situation is being actively conducted.

Speech synthesis technology is applied to many fields such as virtual assistants, audiobooks, automatic interpretation and translation, and virtual voice actors by combining speech recognition technology based on artificial intelligence.

As a conventional speech synthesis method, there are various methods such as unit selection synthesis (USS) and statistic-based parameter synthesis (HMM-based speech synthesis, HTS). The USS method cuts out speech data into phoneme units and stores them, and finds sound pieces suitable for utterance during speech synthesis. The HTS method extracts parameters corresponding to speech characteristics to create a statistical model, and then converts text into speech based on the statistical model. as a way to reconstruct it. However, the conventional voice synthesis method described above has many limitations in synthesizing a natural voice reflecting the speaker's speech style or emotional expression.

Accordingly, recently, a speech synthesis method for synthesizing speech from text based on an artificial neural network is attracting attention.

To provide a method for segmenting a sequence of characters and a speech synthesis system. In addition, it is to provide an artificial intelligence-based speech synthesis technology that can implement a natural voice as if a real speaker is speaking. Another object of the present invention is to provide a highly efficient artificial intelligence-based speech synthesis technology using a small amount of training data.

The technical problem to be solved is not limited to the technical problems as described above, and other technical problems may be inferred.

A method of dividing a sequence of characters composed of a specific natural language according to an aspect includes dividing the sequence based on at least one punctuation mark included in the sequence, thereby generating a first group including a plurality of subsequences. generating; generating a third subsequence by merging the first subsequence and a second subsequence adjacent to the first subsequence when the length of the first subsequence included in the first group is shorter than the first threshold length step; creating a second group by updating the first group based on the third subsequence; generating a plurality of fifth subsequences by dividing the fourth subsequence according to a predetermined criterion when the length of the fourth subsequence included in the second group is longer than a second threshold length; and generating a third group by updating the second group based on the plurality of fifth subsequences.

In the above-described method, the method further includes transmitting a plurality of subsequences included in the third group to a synthesizer.

In the above-described method, the method further includes merging a predetermined text at the end of each of the plurality of subsequences included in the third group, wherein the transmitting includes the subsequence in which the predetermined text is merged. are sent to the synthesizer.

In the above method, the first threshold length means a length shorter than the second threshold length.

In the above-described method, when the sequence includes at least one space of two or more units, modifying the sequence by changing the space of two or more units to a space of one unit; further comprising.

A computer-readable recording medium according to another aspect includes a program for executing the above-described method in a computer.

The speech synthesis system may divide a sequence of characters composed of a specific natural language into subsequences. Also, the speech synthesis system may merge certain text at the end of the subsequence. Accordingly, the speech synthesis system may operate based on the optimal text length, and thus may generate an optimal spectrogram.

1 is a diagram schematically illustrating an operation of a speech synthesis system.
2 is a diagram illustrating an embodiment of a speech synthesis system.
3 is a diagram illustrating an embodiment of outputting a Mel spectrogram through a synthesizer.
4 is a flowchart illustrating an example of dividing a sequence into subsequences.
5 is a diagram for explaining an example in which a speech synthesis system divides a sequence.
FIG. 6 is a diagram for explaining an example in which the speech synthesis system compares the length of a subsequence with a first threshold length and merges adjacent subsequences with each other.
7 is a diagram for explaining an example in which the speech synthesis system compares the length of a subsequence with a second threshold length and divides the subsequence.
8 is a flowchart for explaining an example in which the speech synthesis system performs predetermined processing on a subsequence and transmits it to the synthesizer.
9 is a diagram for explaining an example in which a speech synthesis system merges a predetermined text at the end of a subsequence.

The terms used in the present embodiments are selected as currently widely used general terms as possible while considering the functions in the present embodiments, which may vary depending on the intention or precedent of a person skilled in the art, the emergence of new technology, etc. have. In addition, in certain cases, there are also terms arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the relevant part. Therefore, the terms used in the present embodiments should be defined based on the meaning of the term and the contents throughout the present embodiments, rather than the simple name of the term.

Since the present embodiments may have various changes and may have various forms, some embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the present embodiments to a specific disclosed form, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present embodiments. The terms used herein are used only for description of the embodiments, and are not intended to limit the present embodiments.

Unless otherwise defined, terms used in the present embodiments have the same meanings as commonly understood by those of ordinary skill in the art to which the present embodiments belong. Terms such as those defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in the present embodiments, have an ideal or excessively formal meaning. should not be interpreted.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS [0012] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS [0010] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS [0010] Reference is made to the accompanying drawings, which show by way of illustration specific embodiments in which the present invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the present invention. It should be understood that the various embodiments of the present invention are different but need not be mutually exclusive. For example, certain shapes, structures, and characteristics described herein may be implemented with changes from one embodiment to another without departing from the spirit and scope of the present invention. In addition, it should be understood that the location or arrangement of individual components within each embodiment may be changed without departing from the spirit and scope of the present invention. Accordingly, the following detailed description is not to be taken in a limiting sense, and the scope of the present invention should be taken as encompassing the scope of the claims and all equivalents thereto. In the drawings, like reference numerals refer to the same or similar elements throughout the various aspects.

On the other hand, in the present specification, technical features that are individually described within one drawing may be implemented individually or may be implemented at the same time.

In this specification, “~ unit” may be a hardware component such as a processor or circuit, and/or a software component executed by a hardware component such as a processor.

Hereinafter, various embodiments of the present invention will be described in detail with reference to the accompanying drawings in order to enable those of ordinary skill in the art to easily practice the present invention.

1 is a diagram schematically illustrating an operation of a speech synthesis system.

A speech synthesis device is a device that artificially converts text into human speech.

For example, the speech synthesis system 100 of FIG. 1 may be an artificial neural network-based speech synthesis system. An artificial neural network refers to an overall model that has problem-solving ability by changing the strength of synaptic bonding through learning in which artificial neurons that form a network through synaptic bonding.

The speech synthesis system 100 may be implemented in various types of devices such as a personal computer (PC), a server device, a mobile device, and an embedded device, and as a specific example, a smartphone or tablet that performs speech synthesis using an artificial neural network. It may correspond to a device, an Augmented Reality (AR) device, an Internet of Things (IoT) device, an autonomous vehicle, robotics, a medical device, an e-book terminal, and a navigation device, but is not limited thereto.

Furthermore, the speech synthesis system 100 may correspond to a dedicated hardware accelerator mounted on the above device. Alternatively, the speech synthesis system 100 may be a hardware accelerator such as a neural processing unit (NPU), a tensor processing unit (TPU), or a neural engine, which is a dedicated module for driving an artificial neural network, but is not limited thereto.

Referring to FIG. 1 , the speech synthesis system 100 may receive a text input and specific speaker information. For example, the speech synthesis system 100 may receive “Have a good day!” as shown in FIG. 1 as text input, and may receive “speaker 1” as speaker information input.

“Speaker 1” may correspond to a voice signal or a voice sample indicating a preset speech characteristic of speaker 1. For example, the speaker information may be received from an external device through a communication unit included in the speech synthesis system 100 . Alternatively, the speaker information may be input from a user through the user interface of the speech synthesis system 100 and may be selected from among various speaker information pre-stored in the database of the speech synthesis system 100, but is limited thereto. not.

The speech synthesis system 100 may output a speech based on a text input received as an input and specific speaker information. For example, the speech synthesis system 100 may say “Have a good day!” and “Speaker 1” may be received as inputs, and a voice for “Have a good day!” in which the speech characteristics of speaker 1 are reflected may be output. The speech characteristics of the speaker 1 may include at least one of various factors such as the speaker 1's voice, a rhyme, a pitch, and an emotion. That is, the output voice may be a voice as if the speaker 1 naturally pronounces “Have a good day!”. A detailed operation of the speech synthesis system 100 will be described later with reference to FIGS. 2 to 4 .

2 is a diagram illustrating an embodiment of a speech synthesis system. The speech synthesis system 200 of FIG. 2 may be the same as the speech synthesis system 100 of FIG. 1 .

Referring to FIG. 2 , the speech synthesis system 200 may include a speaker encoder 210 , a synthesizer 220 , and a vocoder 230 . Meanwhile, in the speech synthesis system 200 illustrated in FIG. 2 , only components related to an embodiment are illustrated. Accordingly, it is apparent to those skilled in the art that the speech synthesis system 200 may include other general-purpose components in addition to the components shown in FIG. 2 .

The speech synthesis system 200 of FIG. 2 may receive speaker information and text as inputs and output a speech.

For example, the speaker encoder 210 of the speech synthesis system 200 may receive speaker information as an input and generate a speaker embedding vector. The speaker information may correspond to a speaker's voice signal or a voice sample. The speaker encoder 210 may receive the speaker's voice signal or voice sample, extract the speaker's utterance characteristics, and represent this as an embedding vector.

The speaker's speech characteristics may include at least one of various factors such as speech speed, pause period, pitch, tone, rhyme, intonation, or emotion. That is, the speaker encoder 210 may represent the discontinuous data values included in the speaker information as a vector composed of continuous numbers. For example, the speaker encoder 210 includes a pre-net, a CBHG module, a deep neural network (DNN), a convolutional neural network (CNN), a recurrent neural network (RNN), a long short-term memory network (LSTM), a BRDNN ( A speaker embedding vector may be generated based on at least one or a combination of two or more of various artificial neural network models, such as a Bidirectional Recurrent Deep Neural Network).

For example, the synthesizer 220 of the speech synthesis system 200 may receive text and an embedding vector representing the speaker's utterance characteristics as inputs, and output speech data.

For example, the synthesizer 220 may include a text encoder (not shown) and a decoder (not shown). Meanwhile, it is apparent to those skilled in the art that the synthesizer 220 may further include other general-purpose components in addition to the above-described components.

The embedding vector representing the speech characteristics of the speaker may be generated from the speaker encoder 210 as described above, and the text encoder (not shown) or the decoder (not shown) of the synthesizer 220 receives the speaker encoder 210 from the speaker encoder 210 . An embedding vector representing a speech characteristic may be received.

A text encoder (not shown) of the synthesizer 220 may receive text as an input and generate a text embedding vector. The text may include a sequence of characters in a particular natural language. For example, the sequence of characters may include alphabetic characters, numbers, punctuation marks, or other special characters.

The text encoder (not shown) may separate the inputted text into a unit of a alphabet, a unit of a letter, or a unit of a phoneme, and may input the separated text into the artificial neural network model. For example, a text encoder (not shown) may generate a text embedding vector based on at least one or a combination of two or more of various artificial neural network models such as pre-net, CBHG module, DNN, CNN, RNN, LSTM, BRDNN. .

Alternatively, the text encoder (not shown) may divide the input text into a plurality of short texts and generate a plurality of text embedding vectors for each of the short texts.

A decoder (not shown) of the synthesizer 220 may receive a speaker embedding vector and a text embedding vector from the speaker encoder 210 as inputs. Alternatively, the decoder (not shown) of the synthesizer 220 may receive a speaker embedding vector from the speaker encoder 210 as an input, and receive a text embedding vector from a text encoder (not shown) as an input.

The decoder (not shown) may input the speaker embedding vector and the text embedding vector to the artificial neural network model to generate voice data corresponding to the input text. That is, the decoder (not shown) may generate voice data for the input text in which the speaker's speech characteristics are reflected. For example, the voice data may correspond to a spectrogram or a mel-spectrogram corresponding to the input text, but is not limited thereto. In other words, the spectrogram or Mel spectrogram corresponds to a verbal utterance of a sequence of characters composed of a specific natural language.

A spectrogram is a graph that visualizes the spectrum of a voice signal. The x-axis of the spectrogram represents time and the y-axis represents frequency, and the value of each time frequency can be expressed as a color according to the size of the value. The spectogram may be a result of performing short-time Fourier transform (STFT) on a continuously given speech signal.

STFT is a method in which a voice signal is divided into sections of a certain length and a Fourier transform is applied to each section. At this time, since the result of performing STFT on the speech signal is a complex value, it is possible to take an absolute value to the complex value to lose phase information and to generate a spectrogram including only magnitude information.

On the other hand, the Mel spectrogram is a readjustment of the frequency interval of the spectrogram to the Mel scale. The human auditory organ is more sensitive in the low frequency band than in the high frequency band, and the Mel Scale expresses the relationship between the physical frequency and the frequency perceived by humans by reflecting these characteristics. The Mel spectrogram may be generated by applying a filter bank based on the Mel scale to the spectrogram.

Meanwhile, although not shown in FIG. 2 , the synthesizer 220 may further include an attention module for generating an attention alignment. The attention module is a module for learning which output of a specific time-step of the decoder (not shown) is most related to which output among the outputs of all time-steps of the text encoder (not shown). A higher quality spectrogram or Mel spectrogram can be output by using the attention module.

3 is a diagram illustrating an embodiment of outputting a Mel spectrogram through a synthesizer. The synthesizer 300 of FIG. 3 may be the same as the synthesizer 220 of FIG. 2 .

Referring to FIG. 3 , the synthesizer 300 may receive a list including input texts and speaker embedding vectors corresponding thereto. For example, the synthesizer 300 includes the input text 'first sentence', the corresponding speaker embedding vectors embed_voice1, the input text 'second sentence', and the corresponding speaker embedding vectors embed_voice2 and the input text 'third sentence'. and a list 310 including embed_voice3, which is a speaker embedding vector corresponding thereto, may be received as an input.

The synthesizer 300 may generate as many Mel spectrograms 310 as the number of input texts included in the received list 310 . Referring to FIG. 3 , it can be seen that Mel spectrograms corresponding to input texts of 'first sentence', 'second sentence', and 'third sentence' are generated.

Alternatively, the synthesizer 300 may generate as many Mel spectrograms 320 and attention alignment as the number of input texts together. Although not shown in FIG. 3 , for example, attention alignment corresponding to each input text of 'first sentence', 'second sentence' and 'third sentence' may be additionally generated. Alternatively, the synthesizer 300 may generate a plurality of Mel spectrograms and a plurality of attention alignments for each of the input texts.

Referring back to FIG. 2 , the vocoder 230 of the speech synthesis system 200 may generate speech data output from the synthesizer 220 as actual speech. As described above, the output voice data may be a spectrogram or a Mel spectrogram.

For example, the vocoder 230 may generate the voice data output from the synthesizer 220 as an actual voice signal using Inverse Short-Time Fourier Transform (ISTFT). However, since the spectrogram or the Mel spectrogram does not include phase information, the ISTFT alone cannot completely reconstruct the actual speech signal.

Accordingly, the vocoder 230 may generate the voice data output from the synthesizer 220 as an actual voice signal using, for example, a Griffin-Lim algorithm. The Griffin-Rim algorithm is an algorithm for estimating phase information from magnitude information of a spectrogram or Mel spectrogram.

Alternatively, the vocoder 230 may generate the voice data output from the synthesizer 220 as an actual voice signal based on, for example, a neural vocoder.

A neural vocoder is an artificial neural network model that generates a voice signal by receiving a spectrogram or a Mel spectrogram as an input. A neural vocoder can learn the relationship between a spectrogram or a Mel spectrogram and a voice signal from a large amount of data, and can generate a high-quality real voice signal through this.

A neural vocoder may correspond to a vocoder based on an artificial neural network model such as, but not limited to, WaveNet, Parallel WaveNet, WaveRNN, WaveGlow, or MelGAN.

For example, the WaveNet vocoder consists of several dilated causal convolution layers and is an autoregressive model that uses sequential features between speech samples. For example, the WaveRNN vocoder is an autoregressive model that replaces the multi-layered dilated causal convolution layer of WaveNet with a Gated Recurrent Unit (GRU).

For example, the WaveGlow vocoder can learn to derive a simple distribution such as a Gaussian distribution from the speech data set (x) by using an invertible transform function. After learning, the WaveGlow vocoder can output a voice signal from a Gaussian distribution sample by using the inverse function of the transform function.

As described above with reference to FIGS. 2 and 3 , the synthesizers 220 and 300 generate spectrograms (or Mel spectrograms) using text and speaker embedding vectors. Here, the text may be a sequence of characters composed of a specific natural language.

The synthesizers 220 and 300 include an encoder neural network and an attention-based decoder recurrent neural network. Here, the encoder neural network generates an encoded representation of each of the characters included in the sequence by processing the sequence corresponding to the text. Then, the attention-based decoder neural network processes the decoder input and the encoded representation to generate a single frame of the spectrogram for each decoder input in the sequence input from the encoder neural network.

On the other hand, when the length of the sequence is too long or short, the synthesizers 220 and 300 may not be able to generate high-quality spectrograms (or Mel spectrograms). That is, when the length of the sequence is too long or too short, the attention-based decoder neural network included in the synthesizers 220 and 300 may not be able to generate high-quality spectrograms (or Mel spectrograms).

Accordingly, the speech synthesis systems 100 and 200 according to an embodiment divide a sequence input to the synthesizers 220 and 300 into a plurality of subsequences. Here, the length of each of the divided subsequences has a length optimized for the synthesizers 220 and 300 to generate high-quality spectrograms (or Mel spectrograms).

Hereinafter, examples in which the speech synthesis systems 100 and 200 divide a sequence of characters composed of a specific natural language into a plurality of subsequences will be described with reference to FIGS. 4 to 9 . For example, the module for dividing a sequence into a plurality of subsequences may be a separate module included in the speaker encoder 210 , the synthesizers 220 and 300 , or the speech synthesis systems 100 and 200 .

In addition, hereinafter, the spectrogram and the Mel spectrogram are described as terms that can be used interchangeably. In other words, although it is described as a spectrogram hereinafter, it may be replaced with a Mel spectrogram. Also, in the following, even if it is described as a Mel spectrogram, it may be replaced with a spectrogram.

4 is a flowchart illustrating an example of dividing a sequence into subsequences.

Referring to FIG. 4 , a method of dividing a sequence into subsequences consists of steps processed in time series in the speech synthesis systems 100 and 200 shown in FIGS. 1 and 2 . Therefore, it can be seen that even if the content is omitted below, the content described above with respect to the speech synthesis systems 100 and 200 shown in FIGS. 1 and 2 is also applied to the method of dividing the sequence of FIG. 4 into subsequences. have.

In operation 410, the speech synthesis systems 100 and 200 generate a first group including a plurality of subsequences by dividing the sequence based on at least one punctuation mark included in the sequence.

For example, when any one of predetermined punctuation marks is included in the sequence, the speech synthesis systems 100 and 200 may divide the sequence based on the corresponding punctuation mark. Here, the predetermined punctuation marks may include at least one of ',', '.', '?', '!', ';', '-', and '^'.

Hereinafter, an example in which the speech synthesis systems 100 and 200 divide a sequence based on predetermined punctuation marks will be described with reference to FIG. 5 .

5 is a diagram for explaining an example in which a speech synthesis system divides a sequence.

5 shows an example of a sequence 510 of characters composed of a specific natural language. Also, the sequence 510 includes two types of punctuation marks 521 and 522 .

The speech synthesis systems 100 and 200 identify characters and punctuation marks included in the sequence 510 . Then, the speech synthesis systems 100 and 200 check whether the punctuation marks 521 and 522 included in the sequence 510 are predetermined punctuation marks.

If the punctuation marks 521 and 522 are predetermined punctuation marks, the speech synthesis system 100 or 200 divides the sequence 510 into subsequences 511 and 512 . For example, the punctuation mark '?' and when the punctuation mark '.' is a predetermined punctuation mark, the speech synthesis systems 100 and 200 divide the sequence 510 to generate subsequences 511 and 512 .

The speech synthesis system 100 , 200 generates a first group comprising subsequences 511 , 512 . For example, in the case of the sequence 510 shown in FIG. 5 , a total of two subsequences 511 and 512 are included in the first group.

Referring back to FIG. 4 , in step 420 , when the length of the first subsequence included in the first group is shorter than the first threshold length, in step 420 , the first subsequence and the first subsequence A third subsequence is generated by merging the second subsequence adjacent to .

The speech synthesis systems 100 and 200 compare the length of the subsequence included in the first group with the first threshold length. Then, the speech synthesis systems 100 and 200 merge a subsequence shorter than the first threshold length with a subsequence adjacent thereto. For example, the first threshold length may be predetermined, and may be adjusted according to specifications of the speech synthesis systems 100 and 200 .

Hereinafter, an example in which the speech synthesis systems 100 and 200 compare the length of the subsequence included in the first group with the first threshold length and merge adjacent subsequences will be described with reference to FIG. 6 .

FIG. 6 is a diagram for explaining an example in which a speech synthesis system compares a length of a subsequence with a first threshold length and merges adjacent subsequences with each other.

6 illustrates subsequences 611 and 612 included in the first group. Here, it is assumed that the subsequences 611 and 612 are adjacent to each other. In other words, it is assumed that the subsequence 612 is positioned immediately after the subsequence 611 in the sequence.

The speech synthesis systems 100 and 200 compare the length of the subsequence 611 with a first threshold length. If the length of the subsequence 611 is shorter than the first threshold length, the speech synthesis systems 100 and 200 generate the subsequence 620 by merging the subsequence 611 and the subsequence 612 . Here, merging means connecting the subsequence 612 to the end of the subsequence 611 .

If the length of the subsequence 611 is shorter than the first threshold length, the synthesizers 220 and 300 may not be able to generate the optimal spectrogram. Accordingly, the speech synthesis systems 100 and 200 may improve the quality of the spectrogram generated by the synthesizers 220 and 300 by merging the subsequence 611 and the subsequence 612 .

Referring back to FIG. 4 , in step 430 , the speech synthesis systems 100 and 200 generate the second group by updating the first group based on the third subsequence.

As described above with reference to step 420 , the speech synthesis systems 100 and 200 may selectively merge at least some of the subsequences included in the first group. That is, when the lengths of all subsequences included in the first group are longer than the first threshold length, the subsequences are not merged.

Accordingly, the second group may include a sequence in which some of the subsequences included in the first group are merged, or the subsequences of the first group may be included in the second group as they are.

In step 440, when the length of the fourth subsequence included in the second group is longer than the second threshold length, the speech synthesis system 100 or 200 divides the fourth subsequence according to a predetermined criterion to form a plurality of fifth subsequences. Create subsequences.

The speech synthesis systems 100 and 200 compare the length of the subsequence included in the second group with a second threshold length. Then, the speech synthesis systems 100 and 200 divide the subsequence longer than the second threshold length. For example, the second threshold length may be predetermined, and may be adjusted according to specifications of the speech synthesis systems 100 and 200 . Also, the second threshold length may be set longer than the first threshold length.

Hereinafter, an example in which the speech synthesis systems 100 and 200 compare the length of the subsequence included in the second group with the second threshold length and divide the subsequence will be described with reference to FIG. 7 .

7 is a diagram for explaining an example in which the speech synthesis system compares the length of a subsequence with a second threshold length and divides the subsequence.

7 shows a subsequence 710 included in the second group. The speech synthesis systems 100 and 200 compare the length of the subsequence 710 with a second threshold length. If the length of the subsequence 710 is longer than the second threshold length, the speech synthesis systems 100 and 200 divide the subsequence 710 to generate a plurality of subsequences 721 and 722 . 7 shows that there are a total of two subsequences 721 and 722, but is not limited thereto. In other words, the subsequence 710 may be divided into three or more.

If the length of the subsequence 710 is longer than the second threshold length, the synthesizers 220 and 300 may not be able to generate the optimal spectrogram. Accordingly, the speech synthesis systems 100 and 200 can improve the quality of the spectrogram generated by the synthesizers 220 and 300 by dividing the subsequence 710 .

The criteria by which the subsequence 710 is divided may vary. As an example, the sub-sequence 710 may be divided based on a point where the sub-sequence 710 is breathable when the sub-sequence 710 is uttered. As another example, the subsequence 710 may be divided based on a space included in the subsequence 710 . However, the criterion by which the subsequence 710 is divided is not limited to the above-described examples.

Referring back to FIG. 4 , in operation 450 , the speech synthesis systems 100 and 200 generate the third group by updating the second group based on the plurality of fifth subsequences.

As described above with reference to step 440 , the speech synthesis systems 100 and 200 may selectively divide at least some of the subsequences included in the second group. That is, when the lengths of all subsequences included in the second group are shorter than the second threshold length, the subsequences are not divided.

Accordingly, the third group may include a sequence in which some of the subsequences included in the second group are split, or the subsequences of the second group may be included in the third group as they are.

Although not shown in FIG. 4 , when two or more units of spaces are included in the sequence, the speech synthesis systems 100 and 200 may correct the sequence by changing two or more units of spaces to one unit of space. Specifically, before step 410 is performed, the speech synthesis systems 100 and 200 check the arrangement of each space in the sequence, and when the sequence includes two or more consecutive spaces, it is converted into one unit of space. can be changed

Meanwhile, the speech synthesis systems 100 and 200 may perform a predetermined process on the subsequences before transmitting the subsequences included in the third group to the synthesizers 220 and 300 . An example in which the speech synthesis systems 100 and 200 perform predetermined processing on subsequences will be described with reference to FIGS. 8 and 9 .

8 is a flowchart illustrating an example in which the speech synthesis system performs predetermined processing on a subsequence and transmits the same to the synthesizer.

In operation 810, the speech synthesis systems 100 and 200 merge a predetermined text at the end of each of the plurality of subsequences included in the third group.

The location sensitive attention-based synthesizers 220 and 300 may generate better spectrograms when a certain amount of text is further included at the end of the subsequence. Accordingly, the speech synthesis systems 100 and 200 may merge a predetermined text at the end of the subsequence included in the third group, if necessary.

Hereinafter, an example in which the speech synthesis systems 100 and 200 merge a predetermined text at the end of a subsequence will be described with reference to FIG. 9 .

FIG. 9 is a diagram for explaining an example in which the speech synthesis system merges a predetermined text at the end of a subsequence.

Referring to FIG. 9 , the speech synthesis systems 100 and 200 may generate subsequences 911 and 912 by dividing a sequence 910 . Here, it is assumed that the subsequences 911 and 912 are subsequences included in the third group.

The speech synthesis systems 100 and 200 merge a predetermined text 930 at the end of each of the subsequences 911 and 912 . Although it is illustrated in FIG. 9 that the predetermined text 930 is '.Kanadarama.', the present invention is not limited thereto. That is, the predetermined text 930 may be set in various ways so that the synthesizers 220 and 300 may generate high-quality spectrograms.

Referring back to FIG. 8 , in step 820 , the speech synthesis systems 100 and 200 transmit a subsequence in which a predetermined text is merged to the synthesizers 220 and 300 .

When the predetermined text 930 is merged at the end of the subsequences 911 and 912, the speech synthesis systems 100 and 200 transmit information on the predetermined text 930 together to the synthesizers 220 and 300. do. Accordingly, the synthesizers 220 and 300 may finally generate a spectrogram in which the predetermined text 930 is excluded.

The description of the present specification described above is for illustration, and those of ordinary skill in the art to which the content of this specification belongs will understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be able Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. For example, each component described as a single type may be implemented in a dispersed form, and likewise components described as distributed may be implemented in a combined form.

The scope of the present embodiment is indicated by the claims to be described later rather than the above detailed description, and it should be construed as including all changes or modifications derived from the meaning and scope of the claims and their equivalents.

200: speech synthesis system
210: speaker encoder
220: synthesizer
230: Vocoder

Claims (5)

A method for segmenting a sequence of characters composed of a specific natural language, the method comprising:
generating a first group including a plurality of subsequences by dividing the sequence based on at least one punctuation mark included in the sequence;
generating a third subsequence by merging the first subsequence and a second subsequence adjacent to the first subsequence when the length of the first subsequence included in the first group is shorter than the first threshold length step;
creating a second group by updating the first group based on the third subsequence;
generating a plurality of fifth subsequences by dividing the fourth subsequence according to a predetermined criterion when the length of the fourth subsequence included in the second group is longer than a second threshold length;
generating a third group by updating the second group based on the plurality of fifth subsequences. The method of claim 1,
transmitting the plurality of subsequences included in the third group to a synthesizer. 3. The method of claim 2,
Further comprising; merging a predetermined text at the end of each of the plurality of subsequences included in the third group;
The transmitting may include transmitting the subsequences in which the predetermined text is merged to the synthesizer. The method of claim 1,
The first threshold length means a length shorter than the second threshold length. The method of claim 1,
When the sequence includes at least one space of two or more units, modifying the sequence by changing the space of two or more units to one unit of space.

Images