Notes · Music TechnologyUK · A-Levels

Sampling

Sampling captures a piece of recorded sound and turns it into a playable, editable instrument or compositional element. This topic sets out how audio is digitised - sample rate, bit depth and the Nyquist limit - and the data-rate trade-offs that follow, then how samples are edited and looped without glitches, mapped across a keyboard by pitch, and organised into key and velocity zones. It closes with the creative use of sampling in the technology-based composition.

5 sections~19 min reading time3 competenciesLevel Foundation 1 · Standard 3 · Advanced 1

T·0999 / 16

Exam profile

C2 · Digitise, edit, loop and map samples and choose sample rate and bit depth with their data-rate trade-offsC2 · Organise samples into key and velocity zones for expressive, realistic playbackC4 · Apply Nyquist and quantisation reasoning to sampling problems

Operators:describeexplaincalculatedeterminejustify

basic level

The AS foundation is what sampling is, sample rate and bit depth, and looping a sample.

higher level

The full A-level adds Nyquist, quantisation and data-rate calculation, multisampling and velocity zones.

Depth

Text

Contents · 5 sections

Sampling

Sampling, sample rate and bit depth#

FoundationPearson 9MT0 - sampling

Sampling a continuous waveform

Fig. 1Sampling measures the continuous waveform at regular instants (the marked points). The sample rate must be at least twice the highest frequency (Nyquist) to reproduce it.

To store sound digitally, the continuous analogue waveform is measured (sampled) at regular, closely-spaced instants, and each measurement is rounded to one of a fixed set of numerical values (quantised). Two settings define the quality of this process: the sample rate, how many samples are taken per second, and the bit depth, how many bits are used to store each sample. Together they determine how faithfully the digital version represents the original sound - and how much data it takes.

The sample rate sets the highest frequency that can be captured. The Nyquist theorem states that to reproduce a frequency correctly the sample rate must be at least twice that frequency. To capture the full audible range up to 20 kHz, the sample rate must therefore be at least 40 kHz, which is why the standard rates are 44.1 kHz (CD) and 48 kHz (video), with 96 and 192 kHz used for high-resolution work. If a frequency above half the sample rate (the Nyquist frequency) reaches the converter, it is misread as a false lower frequency - aliasing - so an anti-aliasing filter removes it before conversion.

The bit depth sets how finely each sample's amplitude is measured, and so the dynamic range and the quantisation noise. With n bits there are 2 to the power n possible levels, and the available dynamic range is approximately 6 dB per bit: 16-bit gives about 96 dB (CD quality) and 24-bit about 144 dB. A higher bit depth places the quantisation noise (the small error from rounding each sample) further below the signal, which is why 24-bit is standard for recording - it gives ample headroom and a very low noise floor to work within.

The two settings answer different questions: the sample rate governs the frequency range (via Nyquist), the bit depth governs the dynamic range and noise floor (via the 6 dB-per-bit rule). A common exam confusion is to attribute frequency response to bit depth or dynamic range to sample rate; keeping the two separate - rate for frequencies, depth for levels - is essential. Dither, a tiny amount of noise added when reducing bit depth, smooths quantisation errors on the final render.

For the exam, be able to state the Nyquist theorem (sample rate at least twice the highest frequency), explain aliasing and the anti-aliasing filter, give the 6 dB-per-bit dynamic-range rule (16-bit about 96 dB, 24-bit about 144 dB), and choose sensible settings for a task. The classic decision is 24-bit for recording (headroom, low noise) at 44.1 or 48 kHz, understanding the data-rate cost of higher settings.

f_s \ge 2 f_{max}, \qquad \text{dynamic range} \approx 6.02 n\ \text{dB}, \qquad \text{data rate} = f_s \times n \times \text{channels}

Sampling relationships

Nyquist, the 6 dB-per-bit dynamic range, and the data rate from sample rate, bit depth and channel count.

Common sample rates

Fig. 2Sample rates: at least 40 kHz is needed to capture 20 kHz (Nyquist); 44.1 kHz (CD) and 48 kHz (video) are standard, with 96 and 192 kHz for high-resolution work.

Worked example

Choosing settings and computing the data rate

For a stereo recording session, choose a sample rate and bit depth and justify the trade-off, then calculate the data rate at 44.1 kHz, 24-bit, stereo, and compare it with 16-bit.

Choose settings
Use 44.1 kHz (captures the full 20 kHz audible range by Nyquist, and matches CD delivery) and 24-bit (about 144 dB dynamic range, giving generous headroom and a very low quantisation-noise floor for recording and processing).
Data rate at 24-bit
data rate = fs x bit depth x channels = 44100 x 24 x 2 = 2,116,800 bit/s, about 2.1 Mbit/s (roughly 265 kB/s).
$44100 \times 24 \times 2 = 2{,}116{,}800\ \text{bit/s}$
Compare with 16-bit
At 16-bit: 44100 x 16 x 2 = 1,411,200 bit/s (about 1.4 Mbit/s), so 24-bit costs 1.5 times the data for 48 dB more dynamic range.
Justify
The extra data of 24-bit is worth it while recording and mixing (headroom and low noise); the final master can be dithered down to 16-bit for delivery to save data.

Result: 44.1 kHz / 24-bit stereo captures 20 kHz with ~144 dB range at 2.1 Mbit/s - 1.5x the 16-bit data, justified by the recording headroom.

Typical mistakes

Attributing frequency response to bit depth or dynamic range to sample rate - rate governs frequencies (Nyquist), depth governs levels (6 dB/bit).
Forgetting the anti-aliasing filter, so frequencies above the Nyquist frequency alias to false lower tones.

Active revision

A stereo session is recorded at 48 kHz and 24-bit. State the highest frequency it can capture, its approximate dynamic range, and calculate its data rate in bits per second.

Active recall

Recall the key points — then reveal.

Sources: Pearson Edexcel Level 3 Advanced GCE in Music Technology (9MT0) Specification (Pearson Edexcel)

Editing and looping samples#

StandardPearson 9MT0 - sampling

Matched loop points on a waveform

Fig. 3A seamless loop has its start and end at matching zero-crossings (same level and phase), so the join has no step and no click. A crossfade loop blends any remaining mismatch.

A raw sample almost always needs editing before it is useful: trimming silence from the start and end so it triggers tightly, setting the start point exactly on the transient so it plays in time, and normalising or gain-adjusting the level. A tight, well-trimmed sample that starts precisely on its attack is the foundation of rhythmic sample playback, and sloppy trimming leaves a sample feeling late or loose.

Looping lets a short sample sustain indefinitely - essential for turning a brief recording of a sustained instrument into a playable note that holds as long as a key is pressed. A sustain loop marks a start and end point within the sustained portion of the sample, and playback repeats that section seamlessly until the key is released. The challenge is making the loop join inaudible, because a mismatch at the loop point causes a click or a pumping glitch on every repeat.

The remedies are the same as for editing joins: set the loop start and end at matching zero-crossings so there is no sudden step in level at the seam, and choose points where the waveform's shape and level match on both sides so the join is smooth. Where a perfect match is impossible, a crossfade loop overlaps and blends the end of the loop into its start, hiding any remaining discontinuity - the reliable general solution for sustained, evolving sounds.

Loop length and placement affect the character. A very short loop can introduce an audible pitched buzz or a static, unnatural sustain; a longer loop within a naturally evolving part of the sample preserves more of the instrument's movement. For sounds with vibrato or tremolo, the loop should span a whole number of those cycles so the movement continues smoothly across the seam rather than jumping.

For the exam, be able to describe sample editing (trimming, setting the start on the transient, level) and looping (a sustain loop repeated until key release), and to explain how to make a loop click-free: matched zero-crossing loop points and, where needed, a crossfade loop. This applies the editing principles of the audio-editing topic to the specific problem of seamless sample sustain.

Worked example

Looping a sustained note

A one-second string sample must sustain indefinitely when a key is held. Describe how to set a click-free loop.

Choose the loop region
Select a section within the steady, sustained middle of the sample (after the attack), long enough to preserve the natural movement (spanning whole vibrato cycles if present).
Place the points at zero-crossings
Set the loop start and end at rising zero-crossings so both ends of the loop are at the same zero level and phase, avoiding a step at the seam.
Crossfade if needed
If the tone still shifts slightly across the join, apply a crossfade loop that blends the end into the start, hiding the discontinuity.
Audition held notes
Hold a key and listen for clicks or pumping across several repeats; a matched, crossfaded loop should sustain smoothly.

Result: Loop a steady mid-section between matched zero-crossings, crossfading if necessary, so the note sustains seamlessly for as long as the key is held.

Typical mistakes

Setting loop points away from matching zero-crossings, causing a click or glitch on every loop repeat.
Using a very short loop that sounds static or buzzy instead of looping a longer, naturally evolving section.

Active revision

A one-second recording of a sustained string note must become a playable note that holds indefinitely. Explain how to loop it without an audible click on each repeat.

Active recall

Recall the key points — then reveal.

Sources: Pearson Edexcel Level 3 Advanced GCE in Music Technology (9MT0) Specification (Pearson Edexcel)

Pitch-mapping and multisampling#

StandardPearson 9MT0 - sampling

Pitch-mapping and multisample roots

Fig. 4A sample plays at its original pitch at the root key (60); far from it, resampling doubles or halves the speed. Multisampling adds more roots so every key is played by a nearby sample.

To play a single recorded sample melodically across a keyboard, it is assigned a root key - the note at which it plays back at its original pitch - and pressing other keys transposes it. Transposition by simple resampling changes the playback speed as well as the pitch, so a note played far above the root plays faster (and shorter, and with raised formants), while one far below plays slower and duller. Close to the root the sample sounds natural; far from it, it sounds increasingly unnatural.

This resampling relationship is the same varispeed behaviour as tape: playing a sample an octave above its root doubles its playback speed (and halves its length), an octave below halves the speed. That is fine for a note or two either side of the root, but stretching one sample across the whole keyboard produces obvious artefacts at the extremes - the 'chipmunk' effect high up and a sluggish, dull sound low down.

Multisampling solves this by recording the instrument at several pitches across its range and mapping each recording to a small band of keys around its own root. Because every key then plays a sample recorded close to that pitch, the amount of transposition is small everywhere, and the artefacts of extreme resampling are avoided. A sampled piano, for instance, might use a separate recording every few semitones, giving a realistic instrument across the whole keyboard.

The number of samples is a trade-off between realism and data. More samples (recording every semitone) gives the most natural result but a large instrument; fewer samples (every octave) is smaller but shows more transposition artefacts between roots. A common compromise is a sample every two to four semitones. Modern samplers also crossfade between adjacent samples across their boundary to avoid an audible jump in tone as you move from one root's zone to the next.

For the exam, be able to explain pitch-mapping (a root key, with other keys transposing by resampling, which changes speed and pitch together) and multisampling (several recordings across the range, each mapped to a narrow zone, to keep transposition small and avoid artefacts), and to reason about the realism-versus-data trade-off in choosing how many samples to record.

Worked example

How many samples for a realistic instrument?

A sampled acoustic guitar sounds natural near its root note but chipmunk-like an octave up. Explain the cause and the multisampling fix.

Diagnose
One sample is mapped across the range, so an octave above the root it plays at double speed, raising the formants and shortening it - the chipmunk artefact.
Multisample
Record the guitar at several pitches across its range (for example every three or four semitones) and map each recording to the keys around its own root.
Keep transposition small
Now every key plays a sample recorded within a semitone or two of that pitch, so the transposition (and its speed/formant artefacts) is small everywhere.
Crossfade the zones
Crossfade adjacent samples across their boundaries so the tone does not jump audibly from one root's zone to the next.

Result: The artefact comes from large transposition of one sample; multisampling with several roots and crossfaded zones keeps transposition small and the guitar realistic across the keyboard.

Typical mistakes

Assuming transposing a sample changes only its pitch - simple resampling changes speed (length) and formants too.
Stretching one sample across the whole keyboard and being surprised by artefacts at the extremes.

Active revision

Explain why a realistic sampled piano uses many recordings mapped across the keyboard rather than one sample transposed everywhere.

Active recall

Recall the key points — then reveal.

Sources: Pearson Edexcel Level 3 Advanced GCE in Music Technology (9MT0) Specification (Pearson Edexcel)

Key zones and velocity zones#

StandardPearson 9MT0 - sampling

Velocity layers across the dynamic range

Fig. 5Velocity layers map how hard a key is struck (0-127) to differently-recorded samples, so dynamics change timbre. Crossfading between layers avoids an audible jump.

A sampler organises its samples into zones - regions of the keyboard and of playing dynamics that each trigger a particular sample. A key zone (or key split) maps a sample to a range of keys, so different parts of the keyboard play different sounds: a multisampled instrument uses many key zones across the range, and a split can put a bass sample on the low keys and a lead on the high keys of the same instrument.

Velocity zones (or velocity layers) map different samples to different ranges of key velocity - how hard a key is struck (a value from 0 to 127 in MIDI). A soft press triggers a quietly-recorded sample and a hard press a loudly-recorded one, so the instrument changes timbre with dynamics, not just level. This is essential for realism, because a real instrument played softly and loudly does not merely change in volume - a soft piano note is mellow, a hard one is bright and percussive - and only separate samples per dynamic capture that.

Velocity switching hard-swaps from one layer to the next at set velocity thresholds, which can produce an audible jump between layers; velocity crossfading blends smoothly between adjacent layers as velocity changes, giving a continuous, natural transition through the dynamic range. More layers give a smoother, more realistic dynamic response at the cost of more samples and data - the same realism-versus-data trade-off as multisampling.

Key and velocity zones combine into a two-dimensional map: for each note, several velocity layers; across the keyboard, many key zones. A fully sampled instrument is therefore a grid of samples - a recording for each pitch band and each dynamic layer - which is why detailed sampled instruments are large. Round-robin (alternating between several samples of the same note) further avoids the 'machine-gun' effect of an identical sample repeating on fast repeated notes.

For the exam, be able to describe key zones (samples mapped across the keyboard) and velocity zones/layers (samples mapped to how hard a key is struck, for dynamic realism), explain velocity switching versus crossfading, and reason about how these zones make a sampled instrument expressive. Understanding that velocity should change timbre, not just volume, is the key insight.

Worked example

Velocity layers for a realistic snare

A sampled snare should go from a soft ghost note to a hard rimshot convincingly. Describe how velocity zones achieve this and how to keep it smooth.

Record several dynamics
Record the snare at several dynamics - soft, medium, hard - so each has its true timbre (mellow when soft, bright and cracking when hard), not just a volume change.
Map to velocity layers
Assign the soft sample to low velocities, the medium to mid velocities and the hard to high velocities, so how hard the key is struck selects the right timbre.
Crossfade the layers
Use velocity crossfading between adjacent layers so the timbre transitions smoothly through the dynamic range rather than jumping at each threshold.
Add round-robin
Provide a couple of alternating samples per layer (round-robin) so repeated hits do not sound identically machine-gunned.

Result: Several dynamic recordings mapped to crossfaded velocity layers (with round-robin) make the snare change timbre with playing strength, smoothly from ghost note to rimshot.

Typical mistakes

Thinking velocity only changes volume - velocity layers change timbre (a soft note is mellow, a hard note bright).
Using hard velocity switching where crossfading is needed, causing an audible jump between dynamic layers.

Active revision

Explain how you would use velocity layers to make a sampled snare respond realistically from a soft ghost note to a hard rimshot, and how to avoid an audible jump between layers.

Active recall

Recall the key points — then reveal.

Sources: Pearson Edexcel Level 3 Advanced GCE in Music Technology (9MT0) Specification (Pearson Edexcel)

Sampling in the technology-based composition#

AdvancedPearson 9MT0 - sampling

The creative sampling workflow

Fig. 6Creative sampling: a source is chopped, pitched and stretched, mapped and arranged, then resampled to layer transformations - turning one sound into original musical material.

In the technology-based composition, sampling is not only a way to play realistic instruments but a creative source of material in its own right. A single recorded sound - a spoken phrase, a field recording, a drum break, a snatch of an instrument - can be chopped, rearranged, pitched, stretched and processed into melodies, rhythms and textures that are the substance of the piece. Sampling turns any recorded sound into raw compositional clay.

The core creative technique is chopping and rearranging: a sample is sliced into segments (often at its transients or into equal beats), and the slices are reordered, repeated and re-triggered to create new rhythms and phrases. A drum break can be re-chopped into a new groove; a vocal phrase can be cut into syllables and reassembled into a hook. Mapping the slices across a keyboard or pads lets them be played and sequenced like an instrument.

Pitch and time manipulation (from the earlier topic) transform sampled material further: pitching a slice up or down builds melodies or chords from a single sound, time-stretching fits a sampled loop to the project tempo, and reversing or heavily processing a sample creates textures unrecognisable from the source. Resampling - recording the processed output back in as a new sample and processing it again - layers transformations to build complex, original sounds.

Sampling must be used responsibly and originally. In an assessed composition the material should be the candidate's own or properly cleared, and the marks reward creative transformation and technical control - how imaginatively and skilfully the source is turned into music - not merely dropping in a recognisable loop. The technique is judged by the musical result and the command of the tools, so purposeful, controlled manipulation is what counts.

For the exam and the composition, be able to describe sampling as a creative process: sourcing a sound, chopping and rearranging it, pitching, stretching and processing it, mapping it to be played, and resampling to layer transformations, all in service of an original musical idea. This ties sampling to synthesis, MIDI and effects as the toolkit of technology-based composition.

Worked example

Turning a spoken phrase into music

From a one-bar spoken-word recording, create both a rhythmic and a melodic element for a composition. Describe the techniques.

Chop it
Slice the phrase at its syllables/transients and map the slices across pads or keys so they can be re-triggered in any order and rhythm.
Build a rhythm
Sequence the slices into a new rhythmic pattern, repeating and re-ordering syllables to create a groove that did not exist in the original delivery.
Build a melody
Pitch selected slices to notes of a scale (preserving formants or not, for effect) so the spoken sounds become a melodic hook or chord.
Resample and process
Bounce the arranged result and process it further (filtering, effects), resampling to layer the transformations into a finished, original element.

Result: Chopping, re-sequencing, pitching to a scale and resampling turn one spoken bar into both a rhythmic groove and a melodic hook - sampling as original composition.

Typical mistakes

Dropping in a recognisable loop unchanged instead of transforming the material creatively and controllably.
Ignoring the origin/clearance of sampled material in an assessed composition.

Active revision

Describe how you would turn a single one-bar spoken-word recording into a melodic and rhythmic element for a technology-based composition, naming the techniques used.

Active recall

Recall the key points — then reveal.

Sources: Pearson Edexcel Level 3 Advanced GCE in Music Technology (9MT0) Specification (Pearson Edexcel) · Ofqual - GCE AS and A level qualifications (subject-level conditions and requirements) (Ofqual)

References & sources

Sources

Pearson Edexcel

Pearson Edexcel Level 3 Advanced GCE in Music Technology (9MT0) Specification

Ofqual

Ofqual - GCE AS and A level qualifications (subject-level conditions and requirements)

Sampling

5 sections~19 min reading time3 competenciesLevel Foundation 1 · Standard 3 · Advanced 1

Sampling

Key points

Choosing settings and computing the data rate

Key points

Looping a sustained note

Key points

How many samples for a realistic instrument?

Key points

Velocity layers for a realistic snare

Key points

Turning a spoken phrase into music

Sources

Sampling

Key points

Choosing settings and computing the data rate

Key points

Looping a sustained note

Key points

How many samples for a realistic instrument?

Key points

Velocity layers for a realistic snare

Key points

Turning a spoken phrase into music

Sources