In the spectrogram view, the number displayed at the top-right corner, corresponding to the frequency value, has always a maximum value of 22050.00 Hz, no matter the sampling rate. I would expect it to go up to half the sampling rate, right?
Because the sound is down-graded to mp3 for preview the frequency go to that 22kHz.
When you down-load the sound you get the original quality.
I see what you mean, but I think this is not the point. It is not about sound quality, but a graphical/display issue.
What I mean is that, for instance, I have uploaded a sound at sampling rate = 11 KHz. Thus, the maximum frequency we can hear, and display, is 5500 Hz. However, that number on the corner always reach 22050 Hz. The same happens with sounds at other sampling rates different than 44100 Hz.
I think that the spectral magnitude is well represented. It is only that number in the corner.
From another point of view, for a sound at sampling rate = 8000 Hz, your number in the corner says that there are partials above 4000 Hz, something that is not possible.
Do you know what I mean?
You are right, but it is not a real audio-analyzer, more an impression i guess.
For the ''freak's'' like we are; better use a spectrum analyzer on the lap-top.
You are both right.
If the spectral display is indeed generated from a low quality preview, then the top freq could be 22050hz, even if the original recording woul allow for higher freqs.
If the starting dound had a sample rate of 8000Hz, then the preview window will still have a top freq of 22050Hz, but the display should be blank above ~4000Hz.
Rhis is easy to test: upload some test tones (sine waves of known frequencies, say 1000, 2000, 5000 and 10000Hz). Ideally, record the test tones at different sampling rates. Then upload and check the spectral displays. Moving the mouse over the spectral image shows the frequency at that point.
There is something wrong with the frequency ruler on the freesound spectrogram:
I just posted 8KHz FLAC files and the number in the top right is out by a factor of 6
( e.g. what is actually 250Hz is measured as 1500Hz )
Maybe the x6 problem is specific to 8KHz FLAC ?