Distorted sound for generated Siri voice – cannot get rid of it

Hi everybody,

I’m having a very weird issue. We use some voice files generated by Siri, and while these sound files play fine on my laptop, they are inaudible (very low volume and totally scrambled) when running it from the target system (currently Raspi 4, Stereo output).

I have never experienced such an issue yet, and it seems to be specific to the Siri voice. What I have tried so far, all with NO EFFECT:

  • Change build sample rate to Preserved
  • Change build encoder quality to 100%
  • Change encoder to FADPCM
  • Put the audio file on the same audio track as another audio file which plays fine
  • Record some voice with Audacity, then copy/paste some Siri voice to the end of my recording (still in Audacity)

Especially the latest point is totally confusing my mind. If it is a file format, sampling rate, codec, whatever issue, okay. But HOW DOES THAT SURVIVE COPY/PASTE???

The sample file is here: http://granjow.net/uploads/temp/music-voice-siri.wav When I play it on my PC, all 3 parts are roughly of the same volume. When playing it form the Raspi, the second part is of slightly lower volume (may be speaker related), and the third part is inaudible.

FMOD API 2.02.18

Thanks for any help.

Just tested it on a second Raspi 4 connected to portable speakers, also over stereo jack. Everything works fine, no distortion. So this is probably a weird kind of hardware issue. Will post more once I find out more :slight_smile:

1 Like


Thank you for sharing the solution.

PS: Afair the issue was that the stereo jack was not plugged in correctly, and voice probably cleared itself out because it was the same on both stereo channels!

1 Like


Thanks for confirming.