Singing sound synthesis system, method and device

Abstract

The invention discloses a singing sound synthesis system, which comprises a storage unit, a beat unit, an input unit and a processing unit, wherein the storage unit is used for storing at least one melody; the beat unit is used for indicating a beat; the input unit is used for receiving a plurality of sound signals; and the processing unit is used for processing the sound signals and generating a synthesized singing sound signal. The sound signals are generated by reading aloud or humming by a user according to the melody and the beat, so each sound signal corresponds to the melody and the beat and is processed directly, and the time and the cost of pre-recording a large number of user corpuses are saved, the system resources are saved, and the song synthesis speed is improved; furthermore, the finally obtained synthesized singing sound has tones of users, and the effect is quite lifelike.

Description

technical field [0001] The invention relates to a singing voice synthesis technology, in particular to a singing voice synthesis system, device and method capable of generating realistic singing voices. Background technique [0002] In recent years, with the gradual maturity of the development of information technology, the processing capability of electronic computing devices has also been greatly improved, enabling the realization of many complex applications, one of which is the related technology of speech or singing voice synthesis. Generally speaking, speech synthesis can generally refer to the technology of artificially generating voices close to real people. At present, there are many related applications, such as: virtual singers, electronic pets, singing software, and the simulation combination of composers and singers. The corresponding demand is also increasing day by day. On the traditional architecture, such as figure 1 As shown, the common voice and singing ...

AI Technical Summary

Problems solved by technology

[0004] (1) It takes a long time to record the corpus to establish a corpus, and the corpus requires a huge storage space

[0005] (2) The word derivation program is complex, consumes a lot of system resources, and is prone to word cutting errors

[0006] (3) As far as the Chinese language is concerned, the effect of singing voice synthesis is not good, and it sounds obviously mechanical

[0007] (4) Limited by the pre-recorded corpus, only fixed timbres can be produced. If you want to change the timbre, you must re-record the corpus

[0008] (5) The overall program is complex, and it takes a long time to generate synthetic singing voices, and it is impossible to obtain synthetic singing voices in real time

[0009] Therefore, on the whole, traditional singing voice synthesis methods cannot meet the needs of general users in terms of cost, efficiency, and fluency of synthesized singing voices.

Images