02-04-2010 09:29 PM - last edited on 02-04-2010 09:30 PM
Anyone else achieve duplex audio beside me? I am running it on Bold 9000 OS 4.6.0.294. If you have got duplex audio working, please add to this thread and indicate the model of the phone and OS version.
02-05-2010 03:29 AM
yes, on my 8900 both 4.6.1 and 5.0, did not test it on other models. did not use any codec, just play and record at the same time.
02-07-2010 06:08 PM
02-08-2010 12:33 PM
It's no surprise -- many people here are bound by various NDAs...
02-08-2010 11:39 PM
Unless you are talking about NDAs in certain RIM program. I don't remember signing any NDA when registering here.
02-09-2010 02:37 AM
I meant both RIM (e.g., Alliance Partner NDA) and employer NDAs and Confidentiality Agreements.
02-09-2010 04:35 AM
well, we are an alliance partner. but i am not aware of rim providing us any code at all, would be strange if i had an nda for my own code.
02-15-2010 04:44 AM - last edited on 02-15-2010 10:50 AM
Just to note our success.
We managed to do it on Curve 8900 (4.6.1.310)
Cannot get over the player buffer caused delay
(for VoIP more than 1 is completely unusable).
02-15-2010 10:25 AM
That's odd. My delay is less than 1/10th of a second over wifi. When it is over 3G, the delay is mostly from a 1/3rd of second to almost a second, but that is delayed twice (when the APN is busy or drops down to edge, the delay could be higher). I have a echo server running on my computer. The phone records and sends the audio to the server and the server sends it back immediately. Upon receiving the packets, the phone immediately sends it to the player. I have to test it on the 8900 and see if this still holds true.
02-15-2010 10:48 AM
Well, I agree ("the phone immediately sends it to the player").
But, when does the sound start to play that 'immediately arrived' package ?
We noticed that player buffers that input and plays it out with significant delay.
Did you notice the same ?
Our tests show that buffer is around 60K. So if you (for example) play AMR (32 bytes per frame type 7) the buffer will introduce a delay of
60kby/32by/frame * 20ms/frame = 37000 ms !!
And that's similar to our results.
You can do a similar calc with pcm, again proven empirically.
Have you experienced/circumvented this ?