Multiple frames to process

[AleksandrTulenkov]

If i use model which take X frames to make predictions, ho to setup app to push store and make model prediction on each 16 (for example) frames

[jaygala223]

You can try something like this. It's what I did for my audio transcription app

for audio_frame in audio_frames:
                sound = pydub.AudioSegment(
                            data=audio_frame.to_ndarray().tobytes(),
                            sample_width=audio_frame.format.bytes,
                            frame_rate=audio_frame.sample_rate,
                            channels=len(audio_frame.layout.channels),
                )
                sound_chunk += sound  

            if len(sound_chunk) >= X*1000:
               # some processing