Read along with the three podcasts, or open the full dictionary entry for Decibel for the definition, Arabic meaning, synonyms and example sentences.
Narrator: Today's word is Decibel — noun, pronounced DES-ih-bel, in IPA /ˈdɛsɪbɛl/.
Narrator: A decibel is a unit used to measure the intensity of sound, or more precisely, a unit expressing the ratio of two values of a physical quantity on a logarithmic scale. The symbol is dB. In everyday use, decibels measure how loud something is — a whisper is around 30 dB, ordinary conversation around 60 dB, and a rock concert around 110 dB.
Narrator: The word is a compound: deci — from Latin decimus, meaning a tenth — plus bel, named after Alexander Graham Bell, the Scottish-born inventor associated with the telephone. The bel was the original unit, but it proved too large for practical measurement. So the decibel — one tenth of a bel — became the standard. It is one of the few scientific units named for a person and prefixed with a metric fraction simultaneously.
Narrator: The logarithmic nature of the decibel scale is important to understand. Because human hearing perceives loudness logarithmically, an increase of 10 dB sounds roughly twice as loud to the human ear, but actually represents ten times the acoustic power. An increase of 20 dB is perceived as about four times louder, but represents a hundredfold increase in power.
Narrator: Decibels are also used in electronics, telecommunications, and signal processing — to measure signal gain, signal-to-noise ratios, and power levels. In these contexts, the dB scale works exactly the same way: a ratio expressed on a logarithmic scale.
Narrator: In register, decibel is a technical and scientific term but widely used in everyday speech when discussing noise, music levels, hearing protection, and noise ordinances.
Narrator: Remember: the decibel is the unit that tames the vastness of the hearing range — from the faintest whisper to the roar of a jet engine — into a compact, comprehensible scale.
Speaker A: My neighbour's having building work done and it is incredibly loud. I looked it up and apparently heavy construction can be over 90 decibels — close to the threshold where hearing damage starts.
Speaker B: Yes, 85 decibels is the general occupational health guideline — sustained exposure above that and you risk hearing damage. And as you say, construction easily exceeds that.
Speaker A: The strange thing about the decibel scale is that 90 doesn't feel that much higher than 80, but apparently the power difference is enormous.
Speaker B: That's the logarithmic scale at work. Every 10 decibels is ten times the sound power — so 90 dB is actually ten times the power of 80 dB, even though it only sounds about twice as loud to our ears. Our hearing compresses the range.
Speaker A: Which is why you need the logarithmic scale in the first place — because a linear scale for sound would be unwieldy. The range from a pin drop to a jet engine spans about 14 orders of magnitude.
Speaker B: Exactly. The decibel compresses that into a manageable 0 to 140-ish range. It's an elegant solution to an enormous measurement problem.
Speaker A: Common mistake people make — saying the volume was decibels high, or treating decibels as simply a synonym for loudness. Technically decibels measure intensity relative to a reference point — they're a ratio, not an absolute measure.
Speaker B: Good point. Sound pressure level in dB SPL is the standard measure of loudness for audio. But in electronics, dB might refer to signal power, voltage ratios, or gain — same unit, different reference. Always worth clarifying which dB measurement you mean in a technical context.
Speaker A: And the plural — is it decibels or just decibel? I hear both.
Speaker B: Decibels for the plural — 90 decibels or 90 dB. The symbol dB is never pluralised with an s — you don't write 90 dBs. It's always 90 dB.
Speaker B: Decibel shows up naturally in audio processing, signal analysis, and IoT prompts. Let's look at how to use it precisely.
Speaker A: It's such a technical term — I imagine prompts using it produce much more specific output.
Speaker B: Exactly. First prompt.
Write a Python script using the sounddevice library that monitors ambient noise in real time and logs a warning when the sound level exceeds 85 decibels for more than three consecutive seconds. Display the current decibel level as a rolling average in the terminal and save events to a timestamped CSV log file.
Speaker B: 85 decibels as the threshold anchors the prompt in a real health standard — the AI knows exactly what level to target.
Speaker A: What about audio processing?
Build a Node.js audio normalisation service that reads WAV files and adjusts their peak amplitude to a target level of negative 3 decibels. Log the original peak level in decibels, the applied gain in decibels, and the output peak level for each file. Use the Web Audio API for browser-side preview.
Speaker B: Specifying the target in decibels and asking for gain logged in decibels keeps everything in a consistent unit — no ambiguity about whether you mean linear or logarithmic values.
Speaker A: How about IoT?
Design an IoT noise monitoring system for a smart office using a Raspberry Pi and a USB microphone. Measure sound levels every second in decibels, store readings in an InfluxDB time-series database, and build a Grafana dashboard showing average decibels by hour, peak decibel events, and a daily noise exposure score per desk zone.
Speaker B: Decibels as the primary measurement unit throughout keeps the data pipeline consistent from sensor to dashboard.
Speaker A: What about telecommunications?
Implement a signal quality monitor in a WebRTC application that measures the signal-to-noise ratio of the audio stream in decibels. Display the current SNR in decibels in the UI, colour-code it green above 20 dB, amber between 10 and 20 dB, and red below 10 dB, and log drops below 10 dB with a timestamp.
Speaker B: SNR thresholds in decibels are standard in telecom engineering — using exact dB values in the prompt produces correctly calibrated code.
Speaker A: One more for machine learning?
Pre-process an audio dataset for a speech recognition model by filtering out all recordings where the ambient noise floor exceeds 40 decibels. Compute the noise floor in decibels for each recording using the WebRTC VAD library, flag recordings above the threshold, and generate a data quality report showing the distribution of noise floor levels across the dataset.
Speaker B: 40 decibels as the noise floor threshold is a realistic audio engineering standard — the AI generates plausible filtering logic rather than an arbitrary number.
Speaker A: Using decibels in prompts forces the AI to use real measurement values rather than vague quality adjectives.
Speaker B: Exactly — technical vocabulary produces technical output.
Speaker B: Prompt example:
Decibel — meaning in Arabic & definition More D courses GL Dictionary