Musichifi: an efficient high-fidelity stereo vocoder

Abstract

20 26 20 00 74 07 J an 2 02 6 A B S T R A C T 2 0 2 6 2 0 0 0 7 4 0 7 J a n 2 0 2 6 FI G . 1 IN PU T MU SI C RE PR ES EN TA TI ON 10 2 HI FI V OC OD ER 10 4 BA ND W ID TH EX TE NS IO N MO DU LE 10 6 MO NO -T O- ST ER EO M OD UL E 10 8 MU SI C GE NE RA TI ON S YS TE M 10 0 1 4 2 3 OU TP UT AU DI O 11 2 LO W R ES OL UT IO N AU DI O HI GH RE SO LU TI ON MO NO A UD IO HI GH RE SO LU TI ON ST ER EO A UD IO 5 1 / 9 MUSIC GENERATION SYSTEM 100 HIGH HIGH LOW RESOLUTION RESOLUTION RESOLUTION AUDIO STEREO AUDIO MONO AUDIO 2 3 4 BANDWIDTH INPUT MUSIC MONO-TO- OUTPUT 1 / 9 HIFI VOCODER EXTENSION REPRESENTATION 1 STEREO MODULE 5 AUDIO 104 MODULE 102 108 112 106 FIG. 1 20 26 20 00 74 07 J an 2 02 6 2 0 2 6 2 0 0 0 7 4 0 7 J a n 2 0 2 6 MUSIC GENERATION SYSTEM 100 LOW RESOLUTION R E S O L U T I O N M O N O A U D I O 234 B A N D W I D T H I N P U T M U S I C M O N O - T O - 1 / 9 H I F I V O C O D E R E X T E N S I O N REPRESENTATION 1S T E R E O M O D U L E 5

Images