Um, so in one case the bat signal was given as 40-100 kHz and in another case as 10-120 kHz. So BW would be somewhere between 60-110 kHz. Since some people can hear the bat sonar, it probably does extend down to 10 or 15 kHz. Would you say a 192 kHz recording could be made to cover a BW of 15-111 kHz? That might have reasonable "fidelity". -----Original Message----- From: linux-audio-user-bounces@xxxxxxxxxxxxxxxxxx [mailto:linux-audio-user-bounces@xxxxxxxxxxxxxxxxxx] On Behalf Of Dan Mills Sent: Monday, November 08, 2004 5:34 PM To: linux-audio-user@xxxxxxxxxxxxxxxxxx Subject: Re: [linux-audio-user] recording bats? On Monday 08 November 2004 21:01, Cornell III, Howard M wrote: > Remember, you have to sample at a rate least twice the highest > frequency of interest. The data suggests you need the at least 192 > kHz rate to record 96kHz fundamental signal. Plus the special microphone. A minor niggle but in some circumstances (and this might be one of them) it can matter, you only need twice the Bandwidth of the signal being recorded, NOT Twice the highest frequency! For all practical purposes in audio applications these are equivalent, but there are cases where subsampling a signal is interesting. For example if I have recorded a ultrasonic source using a 192Khz SR and I know that the source is between say 50 & 60 Khz then I can bandpass filter in the DSP and sub sample the output of the BPF at just above 20Khz (say 24K) which reduces the amount of data I need to store WITHOUT losing any information that was present in the original signal. This is kind of cool for playing with sonar where the DSP requirements for signal correlation get really silly if you work at the input rate. Regards, Dan.
