Training method and device of speech synthesis system and speech synthesis method and device
A training method and speech synthesis technology, applied in the field of speech human-computer interaction, can solve the problems of loss of personality of the speaker, difficult to apply to actual scenes, poor sound quality, etc., and achieve the effect of reducing the cost of customization
Pending Publication Date: 2022-04-01
CLOUDMINDS SHANGHAI ROBOTICS CO LTD
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, there are still some serious problems in speech synthesis: for example, the sound quality is very poor, the speech speed and intonation of all speakers are averaged, and the personality of the speakers is lost, so it is difficult to apply to actual scenarios.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
preparation example Construction
[0124] The third aspect of the present invention also provides a speech synthesis method, using the speech synthesis system obtained by any one of the above training methods to perform speech synthesis. Specifically, the method includes:
[0125] Obtain the speaker code of the target speaker;
[0126] Obtain sentences to be used for synthesizing speech;
[0127] According to the end symbol of the sentence, obtain the sentence pattern identification;
[0128] The Chinese character of described sentence is converted into pronunciation unit vector and tone vector without tone;
[0129] According to the length of the pronunciation unit vector, generate stress vectors with all zeros or other default values; or generate corresponding stress vectors according to the stress words in the sentence;
[0130] Taking the pronunciation unit vector, the pitch vector, the stress vector, the sentence pattern identifier and the speaker code as input, they are sent to a trained speech synthes...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention provides a training method and device of a speech synthesis system and a speech synthesis method and device, belongs to the field of speech man-machine interaction, adopts initial and final consonants and tones for independent modeling, and can enable the speech synthesis system to cover a full modeling unit only with fewer training corpora. The method comprises the steps of collecting training voice of a target speaker to form a training audio, and performing text labeling on the training audio; converting the obtained text annotations into pronunciation unit annotations, wherein the pronunciation unit annotations comprise corresponding initial consonants and vowels with tones; disassembling the pronunciation unit mark and converting the pronunciation unit mark into a pronunciation unit vector without tones and a tone vector; and forming a training sample for training the speech synthesis system based on the training audio, the pronunciation unit vector and the tone vector. According to the invention, independent modeling is carried out by adopting initial and final consonants and tones, so that a speech synthesis system can cover a full modeling unit only by using fewer training corpora.
Description
technical field [0001] The invention belongs to the field of voice human-computer interaction, and in particular relates to a training method and device for a voice synthesis system, and a voice synthesis method and device. Background technique [0002] Due to its natural and friendly interaction mode, voice interaction is gradually recognized by people and widely used in various life scenarios, such as intelligent robots, car voice, smart TV and audio smart toys, virtual humans, etc. Speech synthesis technology, TTS technology, represents the speaking ability and expressive ability of smart devices, and is an important technology in the field of voice interaction. [0003] With the development of TTS technology, the use of high-quality training corpus can already achieve speech synthesis effects that are indistinguishable from real people, but currently speech synthesis service providers only provide a few fixed speakers. Although some manufacturers also provide speaker cu...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & AuthorityApplications(China)
IPC IPC(8): G10L13/02G10L13/10G10L17/02G10L17/04G10L17/18G06F40/117G06F40/166
Inventor董天旭
OwnerCLOUDMINDS SHANGHAI ROBOTICS CO LTD