The classroom was noisy.
"Aoi-senpai, can you hear me?"
Yuki's voice was buried in the surrounding noise.
Aoi moved a bit closer. "Lots of noise."
"How can we communicate reliably?"
"Signal detection theory," Aoi started writing in the notebook. "Technology for extracting target signals from noise."
Mira quietly approached and listened carefully. She was good at hearing conversations even in noise.
Aoi explained. "The human brain is a natural matched filter. It matches input with expected signal patterns."
"Matched filter?"
"You multiply the received signal with a known pattern and calculate correlation. If correlation is high, you judge the target signal is present."
The classroom noise continued. Someone's laughter, a chair scraping, distant conversations.
"But," Yuki said, "all sounds mix together, I can't separate them."
"That's the cocktail party effect. Humans are surprisingly good at it, but machines find it difficult."
Mira wrote in her notebook. "Focus creates signal. Attention is filter."
"Attention becomes a filter," Aoi translated. "As Mira says, selective attention extracts signals."
Yuki tried it. Concentrating only on Aoi's voice. The surrounding sounds receded slightly.
"It became easier to hear."
"That's an adaptive filter. The brain automatically learns noise's statistical properties and suppresses them."
Aoi drew a simple diagram. "Think of it like tuning a radio. You adjust the frequency to pick up the station you want, filtering out others."
"But our brains do this automatically?"
"Yes. With remarkable efficiency. Digital signal processing tries to replicate this, but human auditory systems are still superior in many ways."
Mira drew another diagram. Something like a spectrogram waveform.
Aoi supplemented, "If signal and noise are in different frequency bands, filters can separate them. This is frequency-domain filtering."
"But if they're the same frequency?"
"It gets difficult. The signal-to-noise ratio, SNR, becomes important. If signal strength is sufficiently larger than noise, detection is possible."
Yuki asked, "So speaking louder?"
"That's one method. But smarter is encoding. Adding redundancy and error correction to make it noise-resistant."
Mira wrote a new note. "Repeat important parts"
"Repetition," Yuki understood. "Say important things twice."
"Yes. Temporal redundancy. Sending the same information multiple times compensates for noise loss."
The classroom noise quieted slightly. Break time was ending.
Aoi continued, "Communication systems use pilot signals. Sending known signals to estimate channel state."
"Pilot signals?"
"Like 'test, test.' The receiver observes how it distorts and adjusts filters."
Yuki was impressed. "In human conversation too, we confirm 'Can you hear me?'"
"Exactly. Adaptation through feedback."
Aoi added, "In wireless communication, this becomes crucial. The channel changes constantly—people moving, doors opening. Continuous estimation and adaptation are necessary."
"So communication is always fighting against noise?"
"More like dancing with it. Perfect channels don't exist, so we design systems that work despite imperfections."
Mira smiled. Even with few words, she conveyed much. That was also efficient encoding.
"Even full of noise," Yuki said, "there are ways to communicate."
"Yes. Perfect communication channels don't exist. But understanding noise lets us deal with it."
Silence returned to the classroom. Class was starting.
Mira handed over a small note. "Silence is also information"
"Silence is also information," Aoi nodded.
Yuki put away the note. Finding signals in noise. It applied not just to communication, but to life.
Important messages don't lose to noise.