On this concept: Formula

Not sure which lesson to do first? Find your level and get a plan →

CS Thinking · Computational Thinking · Grade 6-8

Audio Representation

Audio representation is the way a computer stores sound as numeric data.

bit rate≈sample rate×bit depth×channels

Quick Answer

Audio representation is the way a computer stores sound as numeric data. A microphone turns sound into a wave, and the computer measures that wave thousands of times a second (the sample rate), storing each measurement with a fixed number of bits (the bit depth). Use it to work out file sizes and bit rates (sample rate × bit depth × channels), and to explain why one recording sounds clear and another muffled.

Why This Matters

It explains why a minute of CD-quality stereo takes about 10 MB, why a phone call can use far fewer samples than music, and what you lose when you lower the quality to save space. Mixing up the settings causes real errors: forgetting that stereo has two channels halves a size estimate, and thinking a higher sample rate makes sound louder confuses pitch with volume.

Intuitive Explanation

Sound is air pressure rising and falling very fast. Draw it as a wavy line over time. A computer cannot store a smooth line, so it measures the height of the wave at regular moments, like photographing a moving swing many times a second. Each measurement is a sample. CD audio takes 44,100 samples every second and stores each as a 16-bit number, which can be any of 216=65,536 levels.

Two settings decide the quality. The sample rate is how often you measure, and it limits the highest pitch you can capture to about half the rate: 44,100 samples per second covers pitches up to about 22,000 Hz, a little beyond human hearing. The bit depth is how finely each measurement is recorded: 8 bits gives only 256 levels, and the rounding adds a faint hiss. Neither setting is volume. Loudness is the size of the wave, which shows up in the sample values themselves.

The data adds up fast. Bit rate = sample rate × bit depth × channels. CD stereo is 44,100 × 16 × 2 = 1,411,200 bits per second, about 176 KB each second and about 10.6 MB a minute. A traditional phone call uses 8,000 samples per second, 8 bits each, one channel: 64,000 bits per second, 22 times less. That is enough for speech but makes music sound thin.

Think of it like this. Sampled sound works like a flip-book: still drawings shown fast enough look like smooth motion. More pages per second (a higher sample rate) catch quicker movements, and more precise drawings (more bits per sample) show finer positions. Adding pages does not make the drawings any bigger, just as a higher sample rate does not make the sound louder.

Core idea. Sound quality depends on how often the wave is sampled and how precisely each sample is stored.


When to Use

Use audio representation when a question mentions sample rate, bit depth, channels (mono or stereo), or the size or bit rate of a sound file. Typical tasks: work out how many bytes a recording needs, explain why one recording sounds better than another, or describe what happens when a sound is digitized. If the question is how MP3 or similar formats shrink the file, that is Data Compression. For pictures instead of sound, use Image Representation.

Tip. Ask: Is a sound being stored as samples taken at a sample rate, with a bit depth and a number of channels?

How to Recognize It

Check whether the numbers describe how often a sound wave is measured and how precisely each measurement is kept.

  1. Are you given samples per second, such as 44,100 or 48,000?

    Yes points to audio representation; that number is the sample rate. If you are given a width and height in pixels instead, it is Image Representation.

  2. Is the answer found by multiplying sample rate, bits per sample, channels and seconds?

    Yes is the raw audio size. If an MP3 or AAC file is making that number smaller by dropping sounds people barely hear, it is Data Compression.

  3. Does the answer need converting between bits and bytes, or into kilobits per second?

    Yes is part of most audio questions, since bit rates are given in bits. If that unit change is the whole question, with no sound involved, it is Bits and Bytes.

  4. Is the question about how the sound is measured, rather than how loud it is?

    Yes keeps you at representation. Loudness is the size of each sample's value, which a higher sample rate does not change.

Audio Representation vs Image Representation vs Data Compression vs Bits and Bytes

Sound files raise questions about measuring, shrinking and counting bits all at once. Ask whether you are describing how sound is sampled, how a picture is stored, how a file is made smaller, or only converting units.

Audio Representation

Meaning
Audio representation stores sound by measuring the sound wave many times a second and keeping each measurement as a number.
Key test
Sample rate, bit depth and channels decide the answer.
Formula
bit rate=sample rate×bit depth×channels
Example
CD audio uses 44,100 samples per second, 16 bits each, in 2 channels: 1,411,200 bits per second.

Image Representation

Meaning
Image representation keeps a picture as rows and columns of pixels, with color numbers for each one.
Key test
Pixels across and down take the place of samples per second.
Formula
pixels=w×h
Example
A 640 by 480 photo has 307,200 pixels.

Data Compression

Meaning
Data compression rewrites a file in fewer bits, either exactly recoverable or with some detail discarded.
Key test
The sampling stays the same, but the file shrinks.
Formula
ratio=originalcompressed
Example
A 1,411 kbps CD track stored as a 128 kbps MP3 is about 11 times smaller.

Bits and Bytes

Meaning
Bits and bytes are units for amounts of data, with 8 bits making 1 byte.
Key test
Only a unit conversion or a count of values is asked.
Formula
1 byte=8 bits
Example
A 128 kilobit-per-second stream delivers 16 kilobytes each second.

Formula & Notation

bit rate≈sample rate×bit depth×channels
Digital audio stores a sequence of sampled amplitude values taken from an analog waveform. The sampling frequency and quantization depth determine fidelity and storage size.

Worked Examples

Example 1 — Size of a voice memo

Easy

A 30-second mono voice memo is recorded at 22,050 samples per second with 16 bits per sample. How many samples are stored, and how big is the file before compression?
Show the steps and the answer
  1. Count the samples: 22,050 per second × 30 seconds = 661,500 samples.

    Mono means one channel, so there is one sample for each moment measured.

  2. Convert the bit depth to bytes: 16 bits = 2 bytes per sample.

    File sizes are counted in bytes, and 8 bits make a byte.

  3. Multiply: 661,500 × 2 = 1,323,000 bytes, about 1.3 MB.

    Every sample takes the same space, so size = samples × bytes per sample.

  4. Check with the bit rate: 22,050 × 16 × 1 = 352,800 bits per second; × 30 s = 10,584,000 bits; ÷ 8 = 1,323,000 bytes.

    Reaching the same number by a second route catches slips in units.

Answer: 661,500 samples, and 1,323,000 bytes (about 1.3 MB).

Takeaway: Size = sample rate × seconds × bytes per sample × channels.

Example 2 — Sample rate sets pitch, not loudness

Standard

A flute note at 3,000 Hz is recorded twice, once at 8,000 samples per second and once at 44,100. Then the flute plays a higher note at 5,000 Hz. Which recordings capture each note correctly? Does the higher sample rate make the recording louder?
Show the steps and the answer
  1. Find each recording's limit, about half the sample rate: 4,000 Hz for 8,000 samples per second, about 22,050 Hz for 44,100.

    To capture a wave you need at least two samples in every up-and-down cycle.

  2. The 3,000 Hz note is below both limits, so both recordings capture it.

    8,000 samples per second gives about 8,000 ÷ 3,000 ≈ 2.7 samples per cycle, just enough.

  3. The 5,000 Hz note is above 4,000 Hz, so only the 44,100 recording captures it correctly.

    With only 8,000 ÷ 5,000 = 1.6 samples per cycle, the measurements trace out a false, lower pitch instead.

  4. Loudness depends on how large the sample values are, and the sample rate does not change that.

    Sample rate controls which pitches are captured, not how loud they are.

Answer: Both recordings capture the 3,000 Hz note; only the 44,100 recording captures the 5,000 Hz note correctly. The higher sample rate does not make the sound louder.

Takeaway: The sample rate sets the highest pitch you can record, about half the rate; volume lives in the size of the samples.

Example 3 — Find the mistake: forgetting the second channel

Application

A 4-minute song is recorded in stereo at 44,100 samples per second, 16 bits per sample. Kofi estimates the size: "44,100 × 2 bytes × 240 seconds = 21,168,000 bytes, about 21 MB." Find the mistake and correct it.
Show the steps and the answer
  1. Check each factor: 44,100 samples per second, 2 bytes per sample, 240 seconds. All three are right.

    Checking factor by factor shows exactly which one is missing.

  2. Spot the missing factor: stereo has 2 channels, left and right, each with its own samples.

    Kofi's calculation counts one channel only, which is the size of a mono recording.

  3. Recompute: 44,100 × 2 bytes × 2 channels × 240 s = 42,336,000 bytes, about 42 MB.

    Two channels means twice the samples, so twice the bytes.

  4. Cross-check with the bit rate: 44,100 × 16 × 2 = 1,411,200 bits per second; × 240 = 338,688,000 bits; ÷ 8 = 42,336,000 bytes.

    The same answer by a second route confirms the correction.

Answer: 42,336,000 bytes, about 42 MB, twice Kofi's estimate, because stereo stores two channels.

Takeaway: Always multiply by the number of channels: stereo doubles the data.

Where You'll See It

  • Recording a voice memo

    Your phone measures the sound wave thousands of times a second and saves each measurement as a number — those samples, replayed in order, recreate your voice; more samples per second means a more faithful recording.

  • Why a high-quality song file is larger

    A 44,100-samples-per-second stereo track at 16 bits per sample needs sample rate x bit depth x channels worth of bits each second — that product is the bit rate, and it is why lossless audio files dwarf low-quality ones.

  • Muffled music on a weak connection

    When your network slows down, a streaming app switches to a copy of the song compressed to a lower bit rate, for example 96 kilobits per second instead of 320. Fewer bits arrive each second, and you hear the trade-off as thinner, muffled sound.

Common Mistakes

  1. The slip: A higher sample rate makes the audio louder.

    Why it's tempting: Both 'more samples' and 'louder' feel like turning something up.

    Instead: Sample rate is HOW OFTEN the wave is measured (affecting which frequencies can be captured), not loudness. Volume comes from the amplitude (size) of each sample, a separate property.

  2. The slip: Stereo audio is stored the same way as mono, just labeled differently.

    Why it's tempting: It still sounds like 'one song,' so it seems like one stream of data.

    Instead: Stereo stores TWO channels (left and right), roughly doubling the data versus mono. Forgetting the channel count makes any bit-rate or file-size estimate too small.

  3. The slip: An MP3 stores exactly the same samples as the CD track it came from, just packed into a smaller file.

    Why it's tempting: It plays the same song and usually sounds the same, so nothing seems to be missing.

    Instead: MP3 is lossy: it throws away sound detail listeners hardly notice, and that detail cannot be restored. CD stereo runs at 1,411,200 bits per second; an MP3 at 128,000 bits per second keeps about one-eleventh of that.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. CD audio uses 44,100 samples per second, 16 bits per sample, and 2 channels. What is its bit rate?

    Hint: Multiply sample rate × bit depth × channels.

  2. How many bytes does 10 seconds of mono audio take at 8,000 samples per second and 8 bits per sample?

    Hint: Mono means 1 channel. Divide bits by 8 to get bytes.

  3. About how many megabytes does a 3-minute song take in uncompressed CD quality (1,411,200 bits per second)? Use 1 MB = 1,000,000 bytes.

    Hint: Convert to bytes per second, then multiply by the number of seconds.

  4. How many different loudness levels can each sample store with a 3-bit depth? With a 16-bit depth?

    Hint: n bits give 2n levels.

  5. A song is 32 MB as WAV and 3 MB as MP3. Both are 3 minutes long, 44.1 kHz, 16-bit stereo. Is the size difference explained by the audio settings or by data compression?

    Hint: Compare the settings of the two files. Which ones differ?

  6. A student says: 'My recording is too quiet, so I'll raise the sample rate from 22,050 to 44,100 samples per second.' Correct the plan.

    Hint: Does sample rate measure how loud a sound is, or how often it is measured?

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Why do CDs use 44,100 samples a second?

People hear sounds up to about 20,000 vibrations per second. To capture a wave you need more than two samples for each vibration, so recording sounds up to 20,000 needs over 40,000 samples a second, and 44,100 leaves a little room to spare. Lower rates lose the highest sounds: phone calls once used 8,000 samples a second, which is why voices sounded flat and the letters s and f were hard to tell apart.

What does bit depth change about the sound?

Bit depth sets how many different levels each sample can take. With 8 bits there are only 256 levels, so quiet passages pick up an audible hiss from rounding. With 16 bits there are 65,536 levels, enough that the rounding sits below what most people can hear. Each extra bit doubles the number of levels and adds about 6 decibels between the loudest sound and the background hiss.

Is a recording at 96,000 samples a second better than a CD?

For listening, almost never noticeably. The extra samples capture sounds above 20,000 vibrations a second, which human ears cannot hear, and the file is more than twice as large. Studios still record at high rates because it leaves room for editing, such as slowing a recording down, before the final mix is saved at a normal rate.

Learning Path

Audio Representation

You are here

You need Data Representation (an agreed rule turns numbers into meaning) and Binary (each sample is stored as a binary number) first. Audio Representation adds sampling: measuring a sound wave at a fixed rate, with a fixed number of bits per measurement, in one or more channels. Next, Data Compression shows how formats like MP3 shrink that data by a factor of about 10.

See Also