Modeling VS Sampling | Which Makes Better Digital Piano Sounds?

When it comes to digital pianos, the main challenge facing engineers is the task at hand to reproduce two things accurately: the sound, (tone) and the feel (touch) of an acoustic piano. When you press an acoustic piano key, it activates a hammer, which then strikes string(s), which, in turn, vibrates and makes a sound. Add to that the complex tonal interactions, such as string resonance, damper resonance, cabinet resonance, and natural reverberation, then you see Digital Piano manufacturers have their work cut out for them reproducing this complex sound. Currently, digital pianos use one or a combination of the methods of Sampling and Digital Modeling to reproduce the sound of an acoustic piano and other musical instruments. We are here today to delve into the question of Modeling VS Sampling...which is better?

WHAT ARE SAMPLES?

A sample is a short audio recording of a musical instrument’s sound, or of any other sound (ocean waves, sirens, wind, etc.).

Samples can also be excerpts from recorded songs. For example, a five-second drum beat from a funk song can be a sample.

In this article, we’ll focus on samples used to reproduce the sound of musical instruments, namely the acoustic piano.

Usually, acoustic piano samples are recorded at multiple velocity levels (multi-samples) so that they naturally respond to the way you play the digital piano's keys, translating touch (from the softest pianissimo to the loudest fortissimo).

The higher the quality of these samples and the technologies used to create (or record) them, the more realistic and accurate the final product will be.

In recent years, technology has become so advanced that high-end digital pianos provide sound that’s nearly indistinguishable from a real piano.

In a professional recording studio, with a perfectly tuned acoustic piano (often a grand piano), manufacturers record each note played at different volumes using an array of high-fidelity microphones.

So even though the process is roughly the same for all manufacturers, all digital pianos sound the same because of a multitude of factors:

1) The individual acoustic piano used to record the sound and the unique condition of that instrument.

2) The recording equipment used to capture the sound.

3) The placement of the mic(s) during the recording has limitless variations.

4) The acoustic environment or room where the sampling process was recorded.

5) The post-production and algorithms used to model complex tonal interactions, such as string resonance, damper resonance, cabinet resonance, natural reverberation, etc.

6) The time length of the samples and the amount of memory dedicated to the samples on a digital piano.

7) The number of touch velocity layers recorded for each note. The more touch velocity layers, the more natural volume transitions and better expressiveness you get.

Generally, more memory means that longer/higher-quality samples with more velocity layers can be stored on a digital piano. But, of course, higher-quality equals a higher cost.

Cheaper digital piano models have less memory, so manufacturers have to take a slightly different approach.

Rather than recording each individual key of an acoustic piano, they record every second or third note and then fill in the space between by stretching the samples using modeling technologies.

Another common practice is, to avoid using too much sampling data, many manufacturers cut off a part of the sample to reduce its size.

For example, when you depress the sustain pedal and play, say, C3 on an acoustic piano, the note will continue to resonate for well over 10 seconds.

Many lower-tier digital pianos will only play back the first 3-5 seconds of the real sampled recording, then loop the that sample so that the same part is played over and over, but with gradually decreasing volume to mimic natural decay.

While this helps to create a perception of a longer decay, it’s not really the same decay that you hear on an acoustic piano, which is much more complex and dynamic than a simple volume decrease. 

The same applies to velocity layers.

If there’s only one or two velocity layers recorded for each note on a digital piano, it becomes very difficult to reproduce the touch and wide dynamic range of an acoustic piano.

For an acoustic piano, depending on how hard or soft you press the keys, the sound changes not only in volume but also in character. So, just making it a matter of simple volume change, is an oversimplification.

Now, if one is a beginner, you may not even notice these subtle nuances, and since manufacturers do try to make those transitions as smooth and realistic as possible it may be perfectly acceptable to ones ears.

However, if you have a more trained ear, this type of sound could become a thing that bothers you because you notice it every time you play the instrument.

WHAT IS PHYSICAL MODELING?

Another interesting technology that has been gaining popularity is called Physical Modeling.

Unlike sampling, which records the actual sound of an acoustic piano at different velocities, physical modeling basically recreates the piano sound from scratch. 

Various modeling techniques and advanced software recreate the physical behavior of the acoustic piano, in which hundreds of elements interact with each other, making up the ultimate “perfectly imperfect” sound that we hear.

While sampling is still the most popular technology in digital pianos today, you’ll hardly find a digital piano that doesn’t use some kind of modeling in conjunction with its samples (e.g., for string resonance, damper resonance, etc.) to further improve the sound and make it sound more natural.

There are also some digital pianos that use purely modeled piano sound with no samples at all.

For example, most Roland high-end digital pianos today feature the new Roland FP-90X sound generator that uses only physical modeling to produce the sound. There are also various VST plugins (Virtual Studio Technology) that provide highly detailed piano modeled sounds and effects.