-
Spectral Frequency Question
I have several voice recordings that I am encoding to mp3 to stream at 16kbps. I was given a sample wav file which encodes to 16kbps mp3 fine and which encodes with a lot of sound artifacts. (wav files below)
(wav file that encodes fine)
https://dl.dropboxusercontent.com/u/45836281/speech_orig.wav(wav file that does not encode well)
https://dl.dropboxusercontent.com/u/45836281/secondsample.wavWhen looking at the spectral frequency display of the two files, the one that encodes fine has a frequency under 4k and the one that does not has a frequency over 10k (images below)
(encodes fine frequency)
https://docs.google.com/file/d/0B8U02wD3-W-OWURkdWhKOXFmNnc/edit?usp=sharing
(encodes not so good frequency)
https://docs.google.com/file/d/0B8U02wD3-W-OeExYR1BoWDNRdHc/edit?usp=sharingI believe I understand why the second file encodes to mp3 with artifacts, because of the high frequency, but what I do not understand is why the second file has such a high frequency. I believe the human voice has a frequency of 300hz to 3400hz (https://en.wikipedia.org/wiki/Voice_frequency) so how is it possible that it has a frequency over 10k?
Also, if the frequency is why the encoding is introducing artifacts, is there a way to record audio such that it is within a certain frequency?
Thanks in advance