Menu
Recent Articles
Pic® Basic
Now keep in mind the idea is not to look for single notes in an audio string, although that may also be possible.
Instead the idea would be to identify a sound if it is played the second time.
Audio is nothing else than changing of frequency and change in amplitude. If you go out from the assumption that 1kHz is 1kHz and 10kHz is 10kHz then the amplitude plays no roll. Remember the pic does not read the analog value of the signal here, the pic pin has a threshold over which is a one, under which is a 0.
In my case the PIC® pin I used as an input is not a schmidt input, so the threshold of the input pin is around 1V. Either type of input can be used.
When audio is played to the pic, the PIC® in other words receives on the input pin an analog value, that either goes over the 1V threshold, or it remains under the 1V threshold.
The volume with which the audio is played to the PIC® pin is important. If the volume of the audio is reduced, then some of the audio peaks that was exceeding the 1V threshold before, now does not exceed it any more. This means the PIC® pin will see less 1's (or high's) than it would with the volume adjusted higher.
It would thus be safe to say that a given audio, played at a specific volume, is going to represent a specific frequency on the PIC® pin.
Any change in the audio frequency, as well as a change in the volume, is going to change the specific frequency on the PIC® pin.
And frequency is something the PIC® can count very easily.
There is however one more thing to consider. Just counting the amount of 1's that is received during a time span is not very secure. An adjustable frequency can so easily be played to the PIC® input pin, and at some frequency the pulse count is going to be the same as that of the original audio, hence a faulty identification.
If a single fixed frequency tone is used as the original audio, then that frequency only should be true.
However, if a tune is played, then the frequency received changes from moment to moment, however long a moment in this case may be, but that would be the key to make the identification secure.
If one then take some short time frequency samples then the picture changes considerably.
For each short time the frequency (and amplitude) has to be the same or a mismatch in comparison will occur. So not only the frequency has to be correct, but also the sequence of the different frequencies has to be in the same order.
An example for use could be a simple door striker that gets opened, it could be a gate or anything else that needs secure access.
An easy every day source of an audio signal could be your cell phone containing a pre-recoded audio or even a tune you generated with the phone's composer.
Remember that the distance you hold the speaker from the mic, the phone's playback volume, and environmental noise could affect the recognition process in this very basic setup.
Anyone mind to calculate the possible access combinations we can have here ?
I'm not going to post the full code, only that what will be of use.
Couple of DIM's. I decided to take 9 samples in sequence
Code:
Dim Dur1 As Word ' Duration 1 Counter Dim Dur2 As Word ' Duration 2 Counter Dim Dur3 As Word ' Duration 3 Counter Dim Dur4 As Word ' Duration 4 Counter Dim Dur5 As Word ' Duration 5 Counter Dim Dur6 As Word ' Duration 6 Counter Dim Dur7 As Word ' Duration 7 Counter Dim Dur8 As Word ' Duration 8 Counter Dim Dur9 As Word ' Duration 9 Counter Dim Durs1 As Word ' Duration 1 Counter Saved Dim Durs2 As Word ' Duration 2 Counter Saved Dim Durs3 As Word ' Duration 3 Counter Saved Dim Durs4 As Word ' Duration 4 Counter Saved Dim Durs5 As Word ' Duration 5 Counter Saved Dim Durs6 As Word ' Duration 6 Counter Saved Dim Durs7 As Word ' Duration 7 Counter Saved Dim Durs8 As Word ' Duration 8 Counter Saved Dim Durs9 As Word ' Duration 9 Counter Saved
At program start, load the saved values -
Code:
Start:
A = 0
Durs1 = ERead 1 ' Remember a Word takes up 4 bytes...
Durs2 = ERead 5
Durs3 = ERead 10
Durs4 = ERead 15
Durs5 = ERead 20
Durs6 = ERead 25
Durs7 = ERead 30
Durs8 = ERead 35
Durs9 = ERead 40
' Wait for the switch it start a learn or sample the audio if it is received
Start1:
If SW = 1 Then A = 1 : LED = 1 : DelayMS 5000 'Set flag A for save. SW is pressed when a learn cycle starts
If Ring = 0 Then LED = 0 : GoTo Start1 'check when a signal received. Ring is the Pic Input Pin for the audio frequency
Dur1 = Counter Ring, 100 ' Get freq count for 100ms
If Dur1 < 50 Then GoTo Start1 ' Filter below 500 Hz freq out in case noise was received, redo from Start1
LED = 0
Dur2 = Counter Ring, 100 ' Get freq count for 100ms
If Dur2 < 50 Then GoTo Start1 ' Filter below 500 Hz freq out in case noise was received, redo from Start1
' Receive 200ms of valid audio over 500Hz before can continue
LED = 1
Dur3 = Counter Ring, 100 ' Get freq count for 100ms
If Dur3 < 15 Then Dur3 = 15 ' If the frequency is too low (ie a pause in the audio) then 150Hz is the dummy value
LED = 0
Dur4 = Counter Ring, 100 ' Get freq count for 100ms
If Dur4 < 15 Then Dur4 = 15 ' If the frequency is too low (ie a pause in the audio) then 150Hz is the dummy value
LED = 1
Dur5 = Counter Ring, 100 ' Get freq count for 100ms
If Dur5 < 15 Then Dur5 = 15 ' If the frequency is too low (ie a pause in the audio) then 150Hz is the dummy value
LED = 0
Dur6 = Counter Ring, 100 ' Get freq count for 100ms
If Dur6 < 15 Then Dur6 = 15 ' If the frequency is too low (ie a pause in the audio) then 150Hz is the dummy value
LED = 1
Dur7 = Counter Ring, 100 ' Get freq count for 100ms
If Dur7 < 15 Then Dur7 = 15 ' If the frequency is too low (ie a pause in the audio) then 150Hz is the dummy value
LED = 0
Dur8 = Counter Ring, 100 ' Get freq count for 100ms
If Dur8 < 15 Then Dur8 = 15 ' If the frequency is too low (ie a pause in the audio) then 150Hz is the dummy value
LED = 1
Dur9 = Counter Ring, 100 ' Get freq count for 100ms
If Dur9 < 15 Then Dur9 = 15 ' If the frequency is too low (ie a pause in the audio) then 150Hz is the dummy value
LED = 0
' 900ms audio sample done -
' Note that the same routine is used to learn as well as sample with. This ensures the most accurate learned values vs comparison
Start2:
LED = 1
If A = 1 Then ' If the Learn button was pressed, then A = 1
EWrite 1, [Dur1] ' save the value of each duration
EWrite 5, [Dur2]
EWrite 10, [Dur3]
EWrite 15, [Dur4]
EWrite 20, [Dur5]
EWrite 25, [Dur6]
EWrite 30, [Dur7]
EWrite 35, [Dur8]
EWrite 40, [Dur9]
A = 0
LED = 0
DelayMS 3000
GoTo Start 'done learning, now go load the new audio values
EndIf
' Not learning, A = 0, so check if these received values is the same as the saved ones -
Chk1:
If Dur1 < Durs1 + 14 And Dur1 > Durs1 - 14 Then GoTo Chk2 'Compare error of + / - 14 counts allowed
GoTo Start1 'false frequency input
Chk2:
If Dur2 < Durs2 + 14 And Dur2 > Durs2 - 14 Then GoTo Chk3
GoTo Start1 'false frequency input
Chk3:
If Dur3 < Durs3 + 14 And Dur3 > Durs3 - 14 Then GoTo Chk4
GoTo Start1 'false frequency input
Chk4:
If Dur4 < Durs4 + 45 And Dur4 > Durs4 - 45 Then GoTo Chk5
GoTo Start1 'false frequency input
Chk5:
If Dur5 < Durs1 + 14 And Dur5 > Durs5 - 14 Then GoTo Chk6
GoTo Start1 'false frequency input
Chk6:
If Dur6 < Durs6 + 14 And Dur6 > Durs6 - 14 Then GoTo Chk7
GoTo Start1 'false frequency input
Chk7:
If Dur7 < Durs7 + 14 And Dur7 > Durs7 - 14 Then GoTo Chk8
GoTo Start1 'false frequency input
Chk8:
If Dur8 < Durs8 + 14 And Dur8 > Durs8 - 14 Then GoTo Chk9
GoTo Start1 'false frequency input
Chk9:
If Dur9 < Durs9 + 14 And Dur9 > Durs9 - 14 Then GoTo Chk10
GoTo Start1 'false frequency input
Chk10: ' All 9 comparisons valid
Rel1 = 1
LED = 0 ' gate striker is opened
DelayMS 800
Rel1 = 0 ' Relay 1 off
GoTo Start
End
If you can improve this principle of audio identification or can come up with a better way of doing it with an 8 bit PIC® microcontroller then please let us know by posting on the forum. It will be most interesting.
Using PDS with SPI GLCD based on ST7565R Controller
Graphic LCDs based on the ST7565 are cheaper then GLCDs with other controllers. SPI requires only four pins. If the circuit