I am trying to use PyAudio in Python for real-time audio processing. I use some standard code for non-blocking streaming using a callback function. However, I notice that there is a delay between the chunks recorded that exceeds the duration of each chunk (i.e. frames dropped I assume).
import timeimport pyaudioRATE = 44100CHUNK_DUR = 1CHANNELS = 1p = pyaudio.PyAudio()prev_callback_time = 0.0def callback(data, frame_count, time_info, status): global prev_callback_time print(f"Time when first sample of input buffer was captured: {time_info['input_buffer_adc_time']}.Expected time: {prev_callback_time + CHUNK_DUR}") delay = prev_callback_time + CHUNK_DUR - time_info['input_buffer_adc_time'] print(f"Delay: {delay} seconds ({delay/CHUNK_DUR * 100} %)") prev_callback_time = time_info['input_buffer_adc_time'] return (data, pyaudio.paContinue)chunk_len = int(CHUNK_DUR * RATE)stream = p.open(format=pyaudio.paInt16, input_device_index=0, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=chunk_len, stream_callback=callback)stream.start_stream()while stream.is_active(): time.sleep(0.5)stream.close()p.terminate()The callback processing time is just ~0.0002 seconds so the delay is not because of that. Even using a chunk duration of 1 second, I can see delays up to 200% of that (i.e. 2 seconds of delay)Is there any way I can check whether I really missed 2 secs of audio?
An interesting observation is that the delay is either ~0% or exactly 200%! That means that it sometimes skips frames that are exactly equal to 2 chunks in length