Al technical title is built by PatSnap Al team. It summarizes the technical point description of the patent document.
A synthesis method and singer's technology, applied in speech analysis, speech recognition, instruments, etc., can solve the problems of data under-fitting, small amount of data, unclear singing voice, etc.
Active Publication Date: 2021-03-09
SICHUAN CHANGHONG ELECTRIC CO LTD
View PDF6 Cites 6 Cited by
Summary
Abstract
Description
Claims
Application Information
AI Technical Summary
This helps you quickly interpret patents by identifying the three key elements:
Problems solved by technology
Method used
Benefits of technology
Problems solved by technology
[0003] Due to the high cost of acquiring the singing database, the small amount of data, and the uneven distribution of different pitches, directly adopting the method of multi-person speech synthesis to achieve multi-singer singing is likely to cause the model to underfit the data and the model parameters to be too average for different singers. Leading to unclear vocal pronunciation and low timbre distinction between singers
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 more
Image
Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
Click on the blue label to locate the original text in one second.
Reading with bidirectional positioning of images and text.
Smart Image
Examples
Experimental program
Comparison scheme
Effect test
Embodiment 1
[0063] The invention provides a multi-singer singing voice synthesis method, which can be realized by a software device or a hardware device. figure 1 The schematic flow chart of the multi-singer singing voice synthesis method that the embodiment of the present invention provides, in this embodiment, the model training phase includes the following steps:
[0064] S11. Analyze the voice data of multiple singers, and extract the phrase features of the data, the phoneme pronunciation duration corresponding to the phrase, and the audio frequency spectrum features corresponding to the phrase;
[0065] Multiple singing voice data sets can be obtained by purchasing from professional data companies and music companies or recording by yourself. The data corresponding to each singer includes a score file, a singing voice audio file corresponding to the score, and a phoneme pronunciation duration file corresponding to lyrics in the score. Preferably, in order to analyze the data more co...
Embodiment 2
[0100] A kind of multi-singer singing voice synthesis device described in the embodiment of the present invention comprises:
[0101] Music score editing module, the device provides a score display and editing interface, and provides an interface for selecting a score, uploading a score, creating a new score, and editing a score, so that users can freely create and modify music works;
[0102] Specifically, the input accepted by this module is a score, and the form of obtaining the score includes: the device provides a score list for the user to select, the user uploads the score to the device, edits the score provided by the device, edits the score uploaded by the user, and creates a new score.
[0103] Optionally, the editing elements of the score include clef, key signature, time signature, key, lyrics and note type. The editing of the above elements directly affects the singing content, pitch, pronunciation duration and overall effect of each phoneme in the synthesi...
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 discloses a multi-singer singing synthesis method, and belongs to the technical field of voice synthesis. The synthesis method comprises two stages of model training and model reasoning,and the model reasoning part is finally deployed in the device. The model training comprises the steps of obtaining singing data of multiple singers, extracting musical sentence features, phoneme pronunciation durations and audio frequency spectrum features, wherein the musical sentence features and the phoneme pronunciation durations are arranged according to a phoneme sequence sequence expandedby lyrics, the lengths and the number of phonemes are kept consistent, and the total frame number of the pronunciation durations is consistent with the total frame number of the corresponding frequency spectrum; generating singer vectors for databases of different singers; and taking the musical sentence features and the singer vectors as the input of the model, and taking the spectrum features and the pronunciation durations as the target joint training model of model fitting. The model adopts an adversarial generative network technology to distinguish timbres and pronunciation characteristics of different singers, and keeps the quality of the synthesized song close to the original sound.
Description
technical field [0001] The present invention relates to the technical field of speech synthesis, and more specifically relates to a multi-singer singing voice synthesis method and device. Background technique [0002] With the gradual improvement of singing voice synthesis technology, virtual idols, singing robots, music education, and music pan-entertainment applications derived from this technology have gradually entered people's lives. higher requirement. Multi-singer vocal synthesis is a singing voice synthesis technology that uses a model to generate multiple different singers' timbres. This technology inputs musical scores and specified singer information to synthesize the singing voice of a specified singer's timbre, thereby realizing the diversity of singing voice synthesis. Multi-person voice synthesis technology has gradually matured, but there are still huge challenges in multi-singer voice synthesis technology and few people in the industry have tried it. [00...
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
Application Date:The date an application was filed.
Publication Date:The date a patent or application was officially published.
First Publication Date:The earliest publication date of a patent with the same application number.
Issue Date:Publication date of the patent grant document.
PCT Entry Date:The Entry date of PCT National Phase.
Estimated Expiry Date:The statutory expiry date of a patent right according to the Patent Law, and it is the longest term of protection that the patent right can achieve without the termination of the patent right due to other reasons(Term extension factor has been taken into account ).
Invalid Date:Actual expiry date is based on effective date or publication date of legal transaction data of invalid patent.
Login to view more 
Login to view more