Issues when trying to record at 96 kHz sample rate

Special Interest Groups Audio Processing
1

Hello,

I know this has been already discussed in the forum but on a differently themed topic, so I thought I should open a new one, hope you don’t mind.

So I’m using Audacity 2.0.5 on Windows XP and version 2.0.3 on Windows 7 (tried it on two different computers) and I’m setting sample rate at 96 kHz in order to try record ultrasounds as much as my microphone can pick them up (Nyquist frequency should thus be 48 kHz, so that I’d hopefully record and view in the spectrogram sounds up to 30 kHz since I already pick them up fine up to 24 kHz using a 48 kHz sample rate).

If you were wondering, yes it has to do with bats and curiosity in general :slight_smile:

However, when I set 96 kHz (or even 88.2 kHz) sample rate, my spectrogram appears to be clearly cut from slightly above 20 kHz, while when I set 48 kHz as sample rate is appears normal (with noise etc) up to 24 kHz frequencies, so I know it is picking up signal.

When I physically hit the microphone with my finger I think it kinda oversaturates with signal and I’d see that this would be the only explanable reason of the full 48kHz peaks with otherwise crystal-clear spectrum (see picture with 5 48kHz peaks, those are 5 hits on mic).

What’d this be caused from? I kinda cheated my mind with the easy answer “sound card doesn’t support sampling above 48kHz” because they both came with the computer and thus are cheap and old stuff.

I scored a cheap 96kHz USB sound card/adapter with 1 IN and 1 OUT 3.5mm jacks, hopefully that will work when I’ll have chance to try it out.

Here are pictures of what it’s happening:

48 kHz sample rate, picking up signal up to 24 kHz just fine IMO




88.2 kHz sample rate, notice that clear cut off at slightly above 20 kHz while it could have picked fine at least up to 24 kHz (magnified the spectrogram on a separate try and the last bits of noise appear at 21.5 kHz)




96 kHz sample rate, probably “fake signal” up to 48 kHz when hitting the microphone 5 times in a row with my finger




Any help and ideas would be appreciated :smiley:

2

When you hit the mic (not a good idea for condenser microphones as it can cause permanent damage to the microphone capsule), if the signal “clips” (goes too high to be represented in digital integer format) then the waveform will be cut off abruptly:
firsttrack000.png
This “clipping” distortion creates extremely high harmonics, right up to the Nyquist frequency.
In this example the sample rate is 192,000 which puts the Nyquist frequency at 96000 Hz, which is much higher than my microphone can detect, but it is the distortion, not the tapping sound, that is creating the extreme high frequencies.
firsttrack001.png
This also gives a clear indication why distortion can be damaging to speakers - it can cause extreme high frequencies at much higher power levels than the “tweeters” are designed to handle.

3

Thanks for your reply. :slight_smile:

However why does setting a sample rate of 88.2 kHz or higher cut any signal above 21.5kHz while using a 48 kHz sample rate records and picks up signal just fine up to 24 kHz?

Is it how I hypothesized (sound card not allowing sampling above 48kHz) ?


Thank you :slight_smile:

4

Analogue to Digital converters (ADC) use “anti-alias” filters to prevent “aliasing” distortion. (http://en.wikipedia.org/wiki/Aliasing)

Audio equipment is often limited, using filters, to cut out frequencies beyond the audio range. This is because ultrasonic frequencies can cause weird modulation effects that can reduce the sound quality in the audible range.

My guess is that the designers of your sound card decided that up to 48 kHz it was adequate to rely on the anti-alias filters to limit the bandwidth, but for higher sample rates they added additional 21.5kHz filtering.