On Thu, Aug 21, 2014 at 5:06 AM, Francisco Webber <[email protected]> wrote: > Chris, > I am not knowledgeable in the area but the limits you describe seem to be > the same limits associated with (human) speech recognition. The way the > brain tackles the signal to noise ratio is by actually understanding what a > certain message is about. Based on this understanding the brain makes a > couple of guesses what a word could mean, coming up with some candidate > words from an incomplete perception. Then the semantically most probable > word is chosen. A possible approach could therefore be to convert the morse > stream into word-SDRs (using cortical.io encoder) and to match the candidate > words semantically.
It is like speech. That is why it's a good field for research. It is hierarchical too. First there are "elements" the dash and dot (called "dit" and "day") The dah being three times longer than the dit. There are spaces based on these lengths too. There is no absolute the timing is relative. Next the elements are used to make letters and signs, an alphabet of about 40 or so symbols The letters are used to make words. Almost like in English but it is more like "texting" but with a few extra symbols available The words are used in a very stylized way with common phrases. The conversations remind me very much of beginning foreign language lessons. With exchanges like "My name is Bill, what is your name? Tom. Nice to meet you Tom. Where are you? I am in San Diago California,...." and so on It is very simple and mostly out of a phrasebook. Although one could transcribe Shakespeare to Morse Code VERY few will deviate away from the common phrases. THIS, I think is what makes CW good for study. We can capture real-world conversations and have a hope of actually "understanding" them and using this understanding to drive recognition. This is because real radio operators walkabout such simple subjects using very simple stock phrases. Example is http://www.naqcc.info/cw_qsos.html So I see a four level hierarchical organization where each level can provide good feedback to the one below. There is a zeroth level but I think that is the encoder and it might use a fixed DSP algorithm. I think the first step is to compute and FFT and then work in the frequency domain. Remember these signals are NOT audio. This is "CW" or unmodulated radio frequency. It is just like the old telegraphs where there also was no audio sent over the wire, just switch on and switch off and a battery supplied DC current when the switch was down. -- Chris Albertson Redondo Beach, California
_______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
