Generate DTMF Tones

Question

I am wondering if anyone has come across a way to generate tones in the iPhone SDK. I am trying to generate DTMF tones, and can't seem to find anything substantial out there. I want to be able to specify how long to play the tone for as well (i.e. to simulate holding the button down as opposed to just pressing it briefly..

I found an open source app called iPhreak. It supposedly generates DTMF tones to fool payphones (I Assure you this is not my intention - my company deals with telephone based Intercom systems). The only problem with that application is that there are files missing from the open source project. Perhaps someone else has gotten this project to work in the past?

If anyone has any idea on where I would look for something like this, I would be very appreciative with my votes :)

Abloom

Answer 1

should be easy enough to generate yourself. given that the hardware can playback a pcm buffer (16bit samples) at 44.1 khz (which it surely can with some library function or the other), you're only left with calculating the waveform:

 const int PLAYBACKFREQ = 44100;
 const float PI2 = 3.14159265359f * 2;

 void generateDTMF(short *buffer, int length, float freq1, float freq2)
 {
      int i;
      short *dest = buffer;
      for(i=0; i<length; i++)
      {
           *(dest++) = (sin(i*(PI2*(PLAYBACKFREQ/freq1))) + sin(i (PI2*(PLAYBACKFREQ/freq2)))) * 16383;
      }
 }

the 16383 is done since I'm using additive synthesis (just adding the sinewaves together). Therefore the max result is -2.0 - 2.0 So after multiplying by 16383 I get more or less the max 16 bit result: -32768 - +32767

EDIT: the 2 frequenties are the frequenties from the wikipedia article the other person who answered linked to. Two unique frequencies make a DTMF sound

Ha

Ok, maybe I am a bit behind the curve on this. Can you give me an example of how you would call that (for example for the #3)? I understand the freq's but don't really get the buffer concept. – Abloom

you create a buffer of sufficient length of a 16 bit signed datatype (I'm not sure how this is done in objective c). You then pick the 2 frequencies belonging to the dtmf tone of 3 (697hz and 1477hz). Call my function with a pointer to the buffer, the length you allocated and then the function will fill it with the waveform of the DTMF tone. This Waveform you'll then have to pass to the iPhone library function which can output the contents of a buffer to the audiohardware. – Ha

Note that the buffer is rendered with 44100 samples/seq. So when playing the waveform, the audio hardware should use the same frequenty or else the DTMF frequencies will be off. Also note, that if you want the waveform to last 10 seconds, then the buffer length should be PLAYBACKFREQ*10. I hope this explains it a bit. More info on PCM (the way a waveform is stored in computer memory) can be found here: en.wikipedia.org/wiki/Pulse-code_modulation – Ha

This does make sense now, although I still have no idea where to start... off to the Docs. Thanks very much. – Abloom

Reinier: I believe that w is 2pi*f/c so wouldn't sin part of formula be sin(i*(PI2*(freq1/PLAYBACKFREQ))? – Apostolic

Answer 2

The easy answer is this:

soundArray = [[NSArray alloc] initWithObjects: 
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-0.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-1.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-2.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-3.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-4.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-5.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-6.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-7.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-8.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-9.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-0.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-pound.caf"] autorelease],
    [[[SoundEffect alloc] initWithContentsOfFile: @"/System/Library/Audio/UISounds/dtmf-star.caf"] autorelease],
              nil];

There you have it. All the sounds of a standard phone keypad, in an array, ready for your enjoyment.

Nutting

I am hoping to generate the tones programatically... so i can play them for any amount of time. this may be a good fallback if I can't get it working right. – Abloom

You can loop these, with varying amounts of work depending on how you chose to play them. Of course, they really are two pure sine waves, and easy to create. However, you still have to create them in some finite length and pass them to a sound player, in which case you'd have to loop those too once the sound player run out of data. Either way, you may have to do the extra work of looping. – Nutting

Will this get you rejected from the App store? – Annmaria

Technically, you are not using any prohibited or private APIs. Also, you didn't have to do anything tricky to get outside of your sandbox. I'd use it as is, and if there is a problem, then go ahead and copy those files into your Bundle and distribute with your app. It won't be a huge hit if you got rejected and had to use that as your plan B. – Nutting

This example has memory leaks. Don't post code with obvious memory leaks. – Hysteroid

@rpetrich: How so? Are you talking about retain counts on array objects being 2 after this snippet? – Nutting

@mahbouds: Ignore retain counts, they are for debugging only. Consider ownership. In your example both the surrounding block own the SoundEffects added to the array, but the surrounding blocks don't have a reference to them, thus a leak. Add autorelease calls to fix this. – Hysteroid

I see what you mean. I'm correcting. Thank you. I don't call this a leak, but setting oneself up for a leak :-) – Nutting

Has anyone actually used this technique? Was it rejected by Apple? – Overtire

hey @mahboudz! what is this SoundEffect class? – Chinquapin

@danik: SoundEffect is a class included in many Apple samples. – Nutting

Answer 3

Swift DTMF Sounds

I was experimenting with generating PCM data and came up with this in Swift. This function will generate an [Float] which are the audio samples. You can play them with AVAudio.

Each DTMF is comprised of a pair of tones, a mark length (250 ms), a space length (250 ms), and of course you need to specify a sample frequency (8000 Hz). Mark and Space are usually around 250 ms for what I would call standard human dialing. The sample frequency is fun to play with, but needs to be twice the highest frequency. For fun you can drop it below that to hear what happens.

public static func generateDTMF(frequency1 frequency1: Float, frequency2: Float, markSpace: MarkSpaceType, sampleRate: Float) -> [Float]
{
    let toneLengthInSamples = 10e-4 * markSpace.0 * sampleRate
    let silenceLengthInSamples = 10e-4 * markSpace.1 * sampleRate

    var sound = [Float](count: Int(toneLengthInSamples + silenceLengthInSamples), repeatedValue: 0)
    let twoPI = 2.0 * Float(M_PI)

    for i in 0 ..< Int(toneLengthInSamples) {
        // Add first tone at half volume
        let sample1 = 0.5 * sin(Float(i) * twoPI / (sampleRate / frequency1));

        // Add second tone at half volume
        let sample2 = 0.5 * sin(Float(i) * twoPI / (sampleRate / frequency2));

        sound[i] = sample1 + sample2
    }

    return sound
}

The full Playground can be downloaded on GitHub.

Intact