I've seen quite a few of these only, however, they all seem to have their share of problems from not going high enough in frequency (looking for 22khz or above) or if they are high enough then they do not have long enough clips (more than two seconds) which makes listening hard. Where can i find a good test that is guaranteed to go over my hearing capacity?
AFAIK foobar2000 will generate any tone you ask it for, and of any duration. Assuming you have at least a 48khz sample rate (really 96k is likely to be preferable), go to 'add location' and type 'tone://22000,10' for a 22kHz 10 second long tone.
where do i type that? i tried a lower frequency in the "filter" filed but nothing happened
File Menu IIRC.
Make sure your hardware is capable of playing these signals without trouble. Depending on your signal, some hardware may resample poorly causing false-positives resulting from aliasing. If this is the case, configure foobar2000's resampler to convert the signal to your soundcard's native sample rate.
This site offers some interesting tests:
http://www.audiocheck.net/audiotests_frequencycheckhigh.php (http://www.audiocheck.net/audiotests_frequencycheckhigh.php)
AFAIK foobar2000 will generate any tone you ask it for, and of any duration. Assuming you have at least a 48khz sample rate (really 96k is likely to be preferable), go to 'add location' and type 'tone://22000,10' for a 22kHz 10 second long tone.
Additionally, the command
sweep://8000-10000,10 will generate a sweep.
This site offers some interesting tests:
http://www.audiocheck.net/audiotests_frequencycheckhigh.php (http://www.audiocheck.net/audiotests_frequencycheckhigh.php)
A good one in terms of length and range, though the voiceover is something of a distraction.
Additionally, the command sweep://8000-10000,10 will generate a sweep.
Linear or logarithmic?
The spectrogram displays a straight slanted line so I assume linear.
the spectrogram per default has a logarithmic scale, so I assume the sweep is logarithmic
The discrepancy of opinions made me check myself. It's logarithmic.
Indeed, my mistake. I find no aesthetic merit in the logarithmic scale, and had forgotten I'd set it to linear the very minute I started to use FB, several years ago. In addition to that, I commonly only use sweep to test a narrow bandwitdh of a few thousand Hz. The graph is then indistinguishable from a straight line.
That means I need to redo my listening tests of my ears' (or my equipment's) upper limit. Around halfway through 16KHz and 18KHz doesn't exactly mean 17KHz, though it's perhaps close enough.
Indeed, my mistake. I find no aesthetic merit in the logarithmic scale, and had forgotten I'd set it to linear the very minute I started to use FB, several years ago. In addition to that, I commonly only use sweep to test a narrow bandwitdh of a few thousand Hz. The graph is then indistinguishable from a straight line.
That means I need to redo my listening tests of my ears' (or my equipment's) upper limit. Around halfway through 16KHz and 18KHz doesn't exactly mean 17KHz, though it's perhaps close enough.
I question using pure sine waves for listening tests on the grounds that spectral masking is a huge influence in the human ear.
Around halfway through 16KHz and 18KHz doesn't exactly mean 17KHz, though it's perhaps close enough.
16, 17 and 18 kHz are roughly a semitone apart, so on the (imaginary) piano they are pretty close
spectral masking is a huge influence in the human ear.
I'm aware of that. Suppose I reach, say, 18KHz, I do not conclude from that that a 16KHz lowpass must sound like crap.