Hello, I’m trying to integrate Gemini Flash 2.0 with audio input into my application.
I noticed that the audio data is downsampled to 16Kbps here.
Given this, would it be okay to downsample the audio on the client side to reduce network overhead?
Would doing so affect the quality of the model’s response?
Also, does “16kbps” correspond to the following audio specifications?
- mono audio channel
- 8 bit depth
- 2kHz sample rate