US 11,972,746 B2
Method and system for hybrid AI-based song construction
Dieter Rein, Berlin (DE); and Jurgen Jaron, Berlin (DE)
Assigned to BELLEVUE INVESTMENTS GMBH & CO. KGAA, (DE)
Filed by BELLEVUE INVESTMENTS GmbH & Co. KGaA, Berlin (DE)
Filed on Sep. 16, 2019, as Appl. No. 16/571,395.
Claims priority of provisional application 62/731,193, filed on Sep. 14, 2018.
Prior Publication US 2020/0090632 A1, Mar. 19, 2020
Int. Cl. G10H 1/00 (2006.01); G06N 5/02 (2023.01); G06N 20/00 (2019.01)
CPC G10H 1/0025 (2013.01) [G06N 5/027 (2013.01); G06N 20/00 (2019.01); G10H 2210/105 (2013.01); G10H 2220/101 (2013.01)] 13 Claims
OG exemplary drawing
 
1. A method of hybrid AI/expert engine-based song generation, wherein is provided an audio loop database containing a plurality of audio loops, each of said audio loops having a plurality of performance parameters associated therewith, comprising the steps of:
(a) requiring a user to select a music style from among a plurality of music styles
(b) requiring the user to select at least one initial song part different from any audio loop in said audio loop database;
(c) using an expert engine to automatically analyze said selected at least one song part to produce a song skeleton structure based on said selected music style and said at least one song part, said song skeleton structure comprising a plurality of empty skeleton song parts, each of said empty skeleton song parts being associated with one or more expert parameters provided by said expert system;
(d) adding said user provided song part to at least one of said empty skeleton song parts in said song skeleton structure;
(e) for each of said skeleton song parts not containing said user provided song part, using an AI system to select a plurality of audio loops from said audio loop database to add to said each of said empty skeleton song parts in said song skeleton structure, said AI system selecting said plurality of said audio loops using at least said plurality of performance parameters associated with each audio loop in said database together with said expert parameters associated with each of said skeleton song parts;
(f) for each of said skeleton song parts not containing said user provided song part, adding said selected audio loops to each of said empty skeleton song part in said song skeleton structure, thereby generating a music work containing at least one of said at least one song parts selected by the user and at least one of said at least one skeleton song parts containing only audio loops added by said AI system, said song skeleton structure comprising a hybrid AI/expert engine-based song; and
(g) performing at least a portion of said hybrid AI/expert engine-based song for the user.