System and method of determining if an information handling system produces one or more audio glitches

1. A system, comprising:

at least one processor; and

a memory medium, coupled to the at least one processor, that stores instructions executable by the at least one processor, which when executed by the at least one processor, cause the system to:

provide, to a plurality of users, a plurality of audio files, wherein at least a first audio file of the plurality of audio files includes at least one audio glitch and at least a second of audio file of the plurality of audio files does not include any audio glitch;

receive a plurality of user responses from the plurality of users that listen to the plurality of audio files;

determine, based at least on the plurality of user responses, a first plurality of portions of the plurality of audio files that include at least one audio glitch;

determine, based at least on the plurality of user responses, a second plurality of portions of the plurality of audio files that do not include the at least one audio glitch;

initialize values of a filter of a convolution neural network for detecting at least one pattern associated with the at least one audio glitch;

determine values of the filter based at least on the first plurality of portions of the plurality of audio files that include the at least one audio glitch and the second plurality of portions of the plurality of audio files;

receive audio produced by an information handling system;

provide the audio produced by the information handling system to the convolution neural network;

convolve the audio produced by the information handling system with the filter;

determine, based at least on data from convolving the audio produced by the information handling system with the filter and output data from the convolution neural network, if the information handling system has produced one or more audio glitches;

if the information handling system has produced one or more audio glitches, provide information indicating that the information handling system has produced the one or more audio glitches; and

if the information handling system has not produced one or more audio glitches, provide information indicating that the information handling system has not produced the one or more audio glitches.

2. The system of claim 1 ,

wherein the instructions further cause the system to perform a pooling operation on the data from convolving the audio produced by the information handling system with the filter; and

wherein, to determine, based at least on the data from convolving the audio produced by the information handling system with the filter and the output data from the convolution neural network, if the information handling system has produced the one or more audio glitches, the instructions further cause the system to determine, based at least on output data from the pooling operation and the output data from the convolution neural network, if the information handling system has produced the one or more audio glitches.

3. The system of claim 2 , wherein the pooling operation includes a max pooling operation, a min pooling operation, or an average pooling operation.

4. The system of claim 1 ,

wherein the instructions further cause the system to determine one or more of weights and biases of the convolution neural network; and

wherein the output data from the convolution neural network is based at least on the one or more of weights and biases of the convolution neural network.

5. The system of claim 4 , wherein, to determine the one or more of weights and biases of the convolution neural network, the instructions further cause the system to determine the one or more of weights and biases of the convolution neural network via a gradient descent process or a backwards propagation process.

6. The system of claim 1 , wherein the filter is configured to determine the at least one pattern that includes a change in magnitude of sound that is above a threshold in a period of time, which indicates at least one of the one or more audio glitches.

7. The system of claim 1 ,

wherein the at least one audio glitch includes a first audio glitch; and

wherein, to determine if the information handling system has produced the one or more audio glitches, the instructions further cause the system to determine, via the convolution neural network, that the audio produced by the information handling system includes a second audio glitch, different from the first audio glitch.

8. A method, comprising:

providing, to a plurality of users, a plurality of audio files, wherein at least a first audio file of the plurality of audio files includes at least one audio glitch and at least a second of audio file of the plurality of audio files does not include any audio glitch;

receiving a plurality of user responses from the plurality of users that listen to the plurality of audio files;

determining, based at least on the plurality of user responses, a first plurality of portions of the plurality of audio files that include at least one audio glitch;

determining, based at least on the plurality of user responses, a second plurality of portions of the plurality of audio files that do not include the at least one audio glitch;

initializing values of a filter of a convolution neural network for detecting at least one pattern associated with the at least one audio glitch;

determining values of the filter based at least on the first plurality of portions of the plurality of audio files that include the at least one audio glitch and the second plurality of portions of the plurality of audio files;

receiving audio produced by an information handling system;

providing the audio produced by the information handling system to the convolution neural network;

convolving the audio produced by the information handling system with the filter;

determining, based at least on data from convolving the audio produced by the information handling system with the filter and output data from the convolution neural network, if the information handling system has produced one or more audio glitches;

if the information handling system has produced one or more audio glitches, providing information indicating that the information handling system has produced the one or more audio glitches; and

if the information handling system has not produced one or more audio glitches, providing information indicating that the information handling system has not produced the one or more audio glitches.

9. The method of claim 8 , further comprising:

performing a pooling operation on the data from convolving the audio produced by the information handling system with the filter;

wherein the determining, based at least on the data from convolving the audio produced by the information handling system with the filter and the output data from the convolution neural network, if the information handling system has produced the one or more audio glitches includes determining, based at least on output data from the pooling operation and the output data from the convolution neural network, if the information handling system has produced the one or more audio glitches.

10. The method of claim 9 , wherein the pooling operation includes a max pooling operation, a min pooling operation, or an average pooling operation.

11. The method of claim 8 , further comprising:

determining one or more of weights and biases of the convolution neural network;

wherein the output data from the convolution neural network is based at least on the one or more of weights and biases of the convolution neural network.

12. The method of claim 11 , wherein the determining the one or more of weights and biases of the convolution neural network includes determining the one or more of weights and biases of the convolution neural network via a gradient descent process or a backwards propagation process.

13. The method of claim 8 , wherein the filter is configured to determine the at least one pattern that includes a change in magnitude of sound that is above a threshold in a period of time, which indicates at least one of the one or more audio glitches.

14. The method of claim 8 ,

wherein the at least one audio glitch includes a first audio glitch; and

wherein the determining if the information handling system has produced the one or more audio glitches includes determining, via the convolution neural network, that the audio produced by the information handling system includes a second audio glitch, different from the first audio glitch.

15. An information handling system, comprising:

at least one processor; and

a memory medium, coupled to the at least one processor, that stores instructions executable by the at least one processor, which when executed by the at least one processor, cause the information handling system to:

produce audio output;

provide the audio output to a convolution neural network, stored by the memory medium, trained to determine if the information handling system has produced one or more audio glitches;

convolve the audio output with a first filter of the convolution neural network to produce first convolved audio output data, wherein the first filter is configured to determine that a change in magnitude of sounds is above a threshold in a period of time;

provide data based at least on the first convolved audio output data to a fully connected neural network of the convolution neural network;

determine, based at least on output from the fully connected neural network, if the information handling system produces the one or more audio glitches;

if the information handling system has produced one or more audio glitches, provide information indicating that the information handling system has produced the one or more audio glitches; and

if the information handling system has not produced one or more audio glitches, provide information indicating that the information handling system has not produced the one or more audio glitches.

16. The information handling system of claim 15 ,

wherein the instructions further cause the information handling system to perform a max pooling operation on the first convolved audio output data; and

wherein, to provide the data based at least on the first convolved audio output data to the fully connected neural network of the convolution neural network, the instructions further cause the information handling system to provide data from the max pooling operation to the fully connected neural network of the convolution neural network.

17. The information handling system of claim 15 , wherein, to determine, based at least on output from the fully connected neural network, if the information handling system has produced one or more audio glitches, the instructions further cause the information handling system to determine if an output value of the fully connected neural network is at or above a threshold value.

18. The information handling system of claim 15 ,

wherein the instructions further cause the information handling system to:

perform a first pooling operation on the first convolved audio output data; and

convolve output from the first pooling operation with a second filter to produce second convolved output data; and

wherein the data based at least on the first convolved audio output data is further based on the second convolved output data.

19. The information handling system of claim 18 ,

wherein the instructions further cause the information handling system to:

perform a second pooling operation on the second convolved output data; and

wherein the data based at least on the first convolved audio output data is further based on output from the second pooling operation.