[amauri]

Hi everybody, Anybody already saw the signal after the voice caught with the MMAPI (JSR 135)? I have developed something similar to the desktop using JavaSound, and worked perfect. Now I am trying to do the same thing in a phone but the signal, at the moment of silence, is "crazy", with values very high (of the same magnitude of the voice signal).

The greatest difficulty is to identify what's silence. By now, I'm just testing in the Sun Toolkit and plotting the signal in Octave. Below follows the codes (part of them) for the capture of speech. Someone know what is the problem? Thanks...

p = Manager.createPlayer("capture://audio?bits=8&rate=8000");
p.realize();
rc = (RecordControl)p.getControl("RecordControl");
output = new ByteArrayOutputStream();
rc.setRecordStream(output);
rc.startRecord();
p.start();

//This occurs when the user stop to speak
rc.stopRecord();
rc.commit();
p.stop();
init.append("Gravando...\n");
recordedSoundArray = output.toByteArray();
output.close();
int n = recordedSoundArray.length;
int[] audioData = new int[n];
for (int i = 0; i < n; i++) {
      audioData[i] = recordedSoundArray[i];
      System.out.println(audioData[i]);
}

[hartti]

Are you saying that the sound recording on Sun WTK causes problems (hmmm... there could be some problems in Sun WTK or with your set-up)?

What if you test on a real phone? Does the "silence" still contain too much noise?

Hartti

[amauri]

Hi, When I play the captured audio it's right, but when i look for the signal (audioData in the below code), it seems noisy. Tomorrow I will go to test in Nokia S60 Emulator. Thanks

[amauri]

Hi,

I tested in Nokia S60 Emulator, with amr encoding and too don't worked (The same problem).

Regards.

[hartti]

Ok,

Sorry for the stupid question, but are you familiar on the amr ecoding file format? So are you looking the data in a correct way?
(and no, I am not an expert on amr file format so google would be more effective than me in getting into details :-)

Hartti