We can always record more video data and extract text from the audio and images if needed. In that case we would need algorithms which require less data, or better hardware to process it.