Hello,
I have been working on a project where I have to mix multiple audio signals of the same source coming from different slave smartphones on one master smartphone in a distributed way. Now I have aligned multiple audio packets (Packet level synchronization) in real-time (still unable to do sample-level synchronization) but when I mix them I get a comb filter effect.
My packet contains 40ms of data at a sampling rate of 48KHz. How can I eliminate this effect?
I am more into ways of making it smooth rather than subtracting delayed signals? Is there any kind of filter somewhat "AntiComb Filter" to make this happen?
Regards,
Khubaib Ahmad
You could try randomizing the phase (or alignment delay) of the audio sampling channels. This could be done on a sample by sample basis. This should destroy the interference pattern. I am not sure what effect this would have on the intelligibility of your audio signal, though.
Hi George Slade, Thank you for your answer. I tried this technique:
I took fft of both signals, added magnitudes of both signals, and then combined this summed magnitude with the phase of the reference signal.
i.e |magnitude| * e^(-j*Θ)
But the output it produced was quite noisy.
Note that adding magnitudes is not the same as adding the complex components of the FFT. I would try randomizing the phase of the components and adding the FFT coefficients of each sampling channel vectorially.
Have a look at this article. It may be relevant to your situation.
https://derby.openrepository.com/bitstream/handle/10545/623149/Dynamic%20DiSP%20for%20Sound%20Reinforcement%20and%20Reproduction%20%28preprint%29.pdf?sequence=1&isAllowed=y
George Slade Thank you for your help. We figured out that adding magnitudes of both signals while using a phase of only one signal not only enhances our voice but also diminshes the spatial texture of voice.
Thank you for the help.
- Badges
- Science topic
- Similar topics
- Control Systems Engineering
- Filtering