audio_process_extern seems to give me 1/2 the data....

[themindfactory]

I am intercepting the decoded mp3 data to do some FFT work with it... however I have noticed it seems to be sending 1/2 the data...

I tried this to test:
void audio_process_extern(int16_t* buff, uint16_t len, bool *continueI2S) {
Serial.printf("a %d %d\n", len, millis());
*continueI2S = true;
}

its sending me len == 1152 every 25ms approx, so that is almost exactly the 44100 sample rate of the mp3 being played, however I would expect 2X that data as its stereo, and I have set forcemono to false, what am I not understanding? did I miss something?

The MP3 is also from a stream online....

Richard.

[schreibfaul1]

Exactly, if the source is a stereo transmitter, the right/left channels are transmitted alternately.
"forcemono" actually only generates a mixed signal which is then transmitted right/left. This is required if there is only one loudspeaker.
The situation is different if the source is mono, in which case half of the samples are available.

[schreibfaul1]

forcemono has no influence, audio_process_extern() provides the raw data of the decoder.

[themindfactory]

So then I am still confused why when streaming a stereo mp3 file it gives me 1/2 the data...

info Channels: 2
info SampleRate: 44100
info BitsPerSample: 16
info BitRate: 128000

My test code:
void audio_process_extern(int16_t* buff, uint16_t len, bool *continueI2S) {
  static int sampleCnt = 0;

  sampleCnt += len;
  EVERYMS(1000) {
    Serial.printf("len %d\n", sampleCnt);
    sampleCnt = 0;
  }
}

which prints out 44100 every second.... would this not be 88200 for stereo?

Thanks!

[themindfactory]

or is the size of buff 2 x len (int16_t), then I understand, I expected len to be the len of the int16_t buffer

[schreibfaul1]

Yes: numSamples = len * audio.getChannels()

[themindfactory]

thanks for the library its working very nicely... :-)