Methods and apparatus to determine audio quality
AssigneeGracenote, Inc.
InventorMarkus Kurt Cremer
Life of this patent
Estimating the term…
Abstract
Methods, apparatus, systems and articles of manufacture are disclosed to determine audio quality. Example apparatus disclosed herein include an equalization (EQ) model query generator to generate a query to a neural network, the query including a representation of a sample of an audio signal. Example apparatus disclosed herein also include an EQ analyzer to access a plurality of equalization settings determined by the neural network based on the query; and compare the equalization settings to an equalization threshold to determine if the audio signal is to be removed from subsequent processing.
Claims20
A tangible, non-transitory computer readable medium comprising instructions, which when executed, cause one or more processors to perform a set of operations comprising:
- transforming at least a portion of an audio signal into a simplified representation of the audio signal;
- inputting the simplified representation into an equalization neural network;
- based on inputting the simplified representation into the equalization neural network, determining equalization settings;
- determining whether the equalization settings satisfy an equalization threshold;
- and
- in response to determining that the equalization settings satisfy the equalization threshold, identifying the audio signal as satisfactory audio quality for subsequent processing.
A computer-implemented method comprising:
- transforming at least a portion of an audio signal into a simplified representation of the audio signal;
- inputting the simplified representation into an equalization neural network;
- based on inputting the simplified representation into the equalization neural network, determining equalization settings;
- determining whether the equalization settings satisfy an equalization threshold;
- and
- in response to determining that the equalization settings satisfy the equalization threshold, identifying the audio signal as satisfactory audio quality for subsequent processing.
A computing device comprising:
- one or more processors;
- and
- a tangible, non-transitory computer readable medium comprising instructions, which when executed, cause the one or more processors to perform a set of operations comprising:
- transforming at least a portion of an audio signal into a simplified representation of the audio signal;
- inputting the simplified representation into an equalization neural network;
- based on inputting the simplified representation into the equalization neural network, determining equalization settings;
- determining whether the equalization settings satisfy an equalization threshold;
- and
- in response to determining that the equalization settings satisfy the equalization threshold, identifying the audio signal as satisfactory audio quality for subsequent processing.
Dependencies are inferred from the claim text.