Error while recording sound in Android, HTC.

[Andrey Grusha]

Hi, folks!

I'm working with android native client and experience a bug while testing my application on HTC device. Incoming voice is ok but outgoing is absent.

in webRtc debug log I receive the message:

E/AudioRecord start() status -38

all the rest going as usual and webrtc reports that peerConnection has established successfully.

I have couple different devices (Samsung s3, Samsung s4, Nexus 5, Lenovo, Fly) but this problem seems to reappear only on HTC device that I have.

When I tried to call to apprtc.appspot.com using latest chrome, I faced the same result: only incoming sound, but no outgoig.

More detailed description of particular device that I use is:

Model: HTC One M7

Android : 4.4.2
Firmware Number: 4.19.401.11
Kernel: 3.4.10-g7fe1bad

Please let me know, if there are any suggestions about this problem

[Ami Fischman]

Can you record audio on that device in other apps? 

--

---
You received this message because you are subscribed to the Google Groups "discuss-webrtc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrt...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Henrik Andreasson]

Have done some work on this device (using Chrome for Android) and found that there seems to be some sort of issue with the audio drivers. Recording audio if only capture is active works and I can render audio if only playout is active. The issue is for full duplex where my tests shows that it only works if the output side is created, opened and started first. If any access to the capture side is done first, one side fails (don't recall which one).

One could do device-specific changes to "fix" this case but I have no direct plans to do so at this stage since it would mess up the code rather much. I have observed the same issue since 4.1.x.

[Andrey Grusha]

So, what sort of "device specific fix" can you suggest?

[Henrik Andreasson]

Well, if you can't modify the client one must rewrite the audio backend to ensure that input is never opened before output for this particular device. It it not trivial and can be broken if HTC does fixes on their side at a later stage. Perhaps it is possible to do some "dummy output" first but I have not tested that. In any case; not something we can prioritize at this stage.

On Wed, Apr 23, 2014 at 11:09 AM, Andrey Grusha <andrew...@gmail.com> wrote:

So, what sort of "device specific fix" can you suggest?

--