Hello,
I am trying to find acoustic round-trip latency of Android smartphones. I have using the technique of where I play a beep and listen to it through the phone and then perform convolution and get time index where we get the peak value.
Issues:
Here the issue is that when we play our beep, the AEC and NS of Android smartphones cancel and attenuate the beep and it cant be seen in recorded data.
Background & Relevant Information:
Most of the applications that measure acoustic latency such as the OboeTester app etc. use VOICE_RECOGNITION (https://developer.android.com/reference/android/media/MediaRecorder.AudioSource#VOICE_RECOGNITION) mode. In this mode, no DSP is performed and we get raw data. But this excludes the latency of DSP algorithms performed by the built-in components.
What I want:
I want to find the round trip latency using VOICE_COMMUNICATION mode. In this mode, all components of AEC, AGC, and NS are activated. But this cancels our beep and we can't get accurate results.
Is there any want to find latency while AEC and NS are working. Looking forward for the solution.
Regards,
Khubaib
Have you considered using a voice signal (or a signal similar to voice) and then use cross-correlation to find the delay? A voice signal will most likely pass through the algorithms used in the phone.
Hi @Jonas. Yes I have considered it. But isn't AEC most likely removing it too..??
Khubaib Ahmad as the phone is designed to let through voice signals, it would be strange if the AEC removes such a signal (but it might of cause modified it, but the cross-correlation is most likely robust enough for that). I know that approaches like this is used in the hearing aid industry.
Let's say I play a voice signal that will from a phone speaker and then capture it with the same phone's mic. If AEC is there, a played signal will be set as a reference signal and it will remove the captured signal. That's what I think.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers