Nicely done! I'd try to not sense a clap based on the peak volume, but how fast the volume rises - a clap is a sharp increase in sound energy, which could help distinguish it. Your analysis of the time envelope is interesting; that's not something I've done with the computer yet, and I'm not positive why there's a difference (possibly resonance in the NXT).
I did want to comment on the tone sensitivity. What it actually measures is the sound energy incident on the sensor over a *wide* range of frequencies, and it can integrate this information in two different ways with repect to frequency. So in one mode ("dB") it counts all frequencies over a wide range (roughly 30 Hz up to 10 kHz) fairly equally, while in the second mode ("dBa") it is biased toward sounds around 2 kHz, with less sensitivity at low frequencies.
Can you use this to sense different frequencies? Yes, if... if you can compare the value returned under dB and dBa modes. For a tone around 2 kHz, both modes should return close to the same value, but for a low tone (200 Hz?), the number returned under dB mode should be significantly higher than the dBa value.
The final kicker? Yes, I get these sort of reposnses... but so far only with "pure" tones, like from a signal generator or tuning fork. If you use the sounds produced by the NXT or RCX, there are far too many overtones and harmonics and distortions for this scheme to work... and I've been unable to make it work by "singing" tones to the NXT by mouth either. But, the field of tonal sensing for the NXT is still young...
-- Brian "short post? Riiiight..." Davis
Anonymous said…
Very interesting. Thanks a lot for the clarification. I'll try that out.
For remotely controlling the NXT, good recognition of pure tones may be enough. Perhaps a good whisle or something of this sort would work.
Comments
I did want to comment on the tone sensitivity. What it actually measures is the sound energy incident on the sensor over a *wide* range of frequencies, and it can integrate this information in two different ways with repect to frequency. So in one mode ("dB") it counts all frequencies over a wide range (roughly 30 Hz up to 10 kHz) fairly equally, while in the second mode ("dBa") it is biased toward sounds around 2 kHz, with less sensitivity at low frequencies.
Can you use this to sense different frequencies? Yes, if... if you can compare the value returned under dB and dBa modes. For a tone around 2 kHz, both modes should return close to the same value, but for a low tone (200 Hz?), the number returned under dB mode should be significantly higher than the dBa value.
The final kicker? Yes, I get these sort of reposnses... but so far only with "pure" tones, like from a signal generator or tuning fork. If you use the sounds produced by the NXT or RCX, there are far too many overtones and harmonics and distortions for this scheme to work... and I've been unable to make it work by "singing" tones to the NXT by mouth either. But, the field of tonal sensing for the NXT is still young...
--
Brian "short post? Riiiight..." Davis
For remotely controlling the NXT, good recognition of pure tones may be enough. Perhaps a good whisle or something of this sort would work.