Question

Why do we clip signal values before writing them to the PyAudio output stream?

Answer 1

In Pyaudio output stream, we are using CHUNKS of data, instead of a continuous amount of audio because of processing power. (Let's assume this was a continuous flow of data being received from a Microphone and is being recorded and saved) then it would just eat up the processor, thus causing potential crashes. In terms of Raspberry Pi / Arduino development (Where the RAM is very small) CHUNKING the data like this makes the stream flow easier and thus prevents memory leaks.

Moreover it is very difficult to implement a algorithm for determining for speech or noise by using constant stream flow of sound data.Therefore, by storing this into an array, you can perform analysis on this data, then you couls have some threshold to determine if you want to keep the data aand clip the values for above the threshold.