| CPC H04N 21/8113 (2013.01) [G10H 2210/295 (2013.01)] | 117 Claims |
|
1. A method comprising:
receiving input corresponding to an original audio file and a plurality of audio generators;
training a labeling algorithm using a labeling training dataset, wherein the labeling algorithm is trained to associate audio files with audio labels;
determining audio profiles of a plurality of component audio files associated with the original audio file and frequency profiles of the plurality of audio generators, wherein determining the audio profiles includes applying the trained labeling algorithm to the plurality of component audio files associated with the original audio file to identify audio labels corresponding to the audio profiles for the plurality of component audio files;
determining a routing scheme for the plurality of component audio files and the plurality of audio generators, wherein the routing scheme matches a component audio file to an audio generator based on the audio profiles and the frequency profiles; and
generating an audio output based on the routing scheme, wherein the audio output is generated in accordance with the component audio file routed to the matching audio generator.
|
|
40. A method comprising:
receiving input corresponding to an original audio file and a plurality of audio generators;
determining audio profiles of a plurality of component audio files associated with the original audio file and frequency profiles of the plurality of audio generators;
training a routing algorithm using a routing training dataset, wherein the routing algorithm is trained to associate audio files with audio generators;
determining a routing scheme for the plurality of component audio files and the plurality of audio generators by applying the trained routing algorithm to the plurality of component audio files and the plurality of audio generators to identify the routing scheme for the plurality of component audio files and the plurality of audio generators, wherein the routing scheme matches a component audio file to an audio generator based on the audio profiles and the frequency profiles; and
generating an audio output based on the routing scheme, wherein the audio output is generated in accordance with the component audio file routed to the matching audio generator.
|
|
79. A method comprising:
receiving input corresponding to an original audio file and a plurality of audio generators;
identifying a plurality of component audio files associated with the original audio file by:
training a demixing algorithm using a demixing training dataset, wherein the demixing algorithm is trained to associate audio files with associated pluralities of component audio files; and
applying the demixing algorithm to the original audio file to identify the associated plurality of component audio files;
determining audio profiles of the plurality of component audio files associated with the original audio file and frequency profiles of the plurality of audio generators;
determining a routing scheme for the plurality of component audio files and the plurality of audio generators, wherein the routing scheme matches a component audio file to an audio generator based on the audio profiles and the frequency profiles; and
generating an audio output based on the routing scheme, wherein the audio output is generated in accordance with the component audio file routed to the matching audio generator.
|