Say: I'm just to press the keys on the piano, the software can recognize what I press notes, what are the way and principle of this recognition, the great god give trouble about it

CodePudding user response:

Sound feature depends on the frequency
Sound depends on loudness

You came out of the original sampling data recording for PCM data (PCM) is what you can baidu

Loudness of PCM is actually a continuous time series, the fast Fourier transform can be used and the like into frequency domain information, and then use MFCC algorithm for feature
And then can be compared to a feature similarity

Above is normal audio recognition principle,

The Google open a set of characteristics of the natural sound of millions of data to a library, allows you to do training, recognition and artificial intelligence (download path forget yourself to baidu, this I just amateur level, so not to collect this kind of thing)

CodePudding user response:

So rather than audio recognition, in fact than "pattern recognition", he put the sound information domain into frequency domain graph, then compared the frequency domain graph similarity

CodePudding user response:

reference 1st floor wanghui0380 response:

sound characteristic depends on the frequencySound depends on loudness

You came out of the original sampling data recording for PCM data (PCM) is what you can baidu

Loudness of PCM is actually a continuous time series, the fast Fourier transform can be used and the like into frequency domain information, and then use MFCC algorithm for feature
And then can be compared to a feature similarity

Above is normal audio recognition principle,

The Google open a set of characteristics of the natural sound of millions of data to a library, allows you to do training, recognition and artificial intelligence (download path forget yourself to baidu, this I just amateur level, so not to collect this kind of thing)

thanks to the great god give advice or comments

CodePudding user response:

I probably understand this principle, thanks to the great god grant instruction. Thank you very much...

CodePudding user response:

If your software is a microphone to collect voice to identify notes that's the principle of the # 1 said,
If your instrument is a MIDI device, then your software may be collected is the MIDI signal, MIDI device is an external input and output devices, and communication with the host software according to the MIDI specification, which one do you press the button, it will send corresponding signal, host software, you will know what notes,

CodePudding user response:

Thanks to the great god reply, MIDI access way seems to be simple,