Multisamples vs ordinary samples

[stevewaits]

Hello,

I find this section of HC forums the most appropriate for this question. So here it goes:

 

Which do you prefer and why? Do multisamples really show the "character" of an instrument as they are in the most cases recorded key by key? Let's stick to synthesizers here unless you have a good reason not to Or do you prefer traditional samples as they are more lighteight and there is no problem with tuning them?

 

3.. 2... 1... Start! Samples vs multisamples - what are your thoughts?

[Phil O'Keefe]

I'm usually looking for realism and fidelity to the source, so I tend to prefer multisamples. Single samples, even if done on every note of the keyboard, just can't match the dynamic and tonal realism of stacked, velocity-switched multisamples.

[Etienne Rambert]

Hello,

I find this section of HC forums the most appropriate for this question. So here it goes:

 

Which do you prefer and why? Do multisamples really show the "character" of an instrument as they are in the most cases recorded key by key? Let's stick to synthesizers here unless you have a good reason not to Or do you prefer traditional samples as they are more lighteight and there is no problem with tuning them?

 

3.. 2... 1... Start! Samples vs multisamples - what are your thoughts?

 

 

Depends on the needs of the music I'm recording. Sometimes a canned wavetable patch fits better than a behemoth, RAM-devouring multi-sampled patch. At other times, more realism is needed.

[ChipCurtis]

If you're going to sample, say, an acoustic piano... you MUST multisample. Because a low note on a piano sounds NOTHING like a high note on a piano. Completely different timbral character, and you can't just take a single piano note and transpose it across the 88 keys.

 

For something like synthesizer waveforms, a single-cycle of that waveform is all that's needed, and can then be transposed up and down the keyboard for pitch. The synth's filter, envelopes, and other components are what give the synth sound its character... not just the waveform sample itself.

 

Of course, single-cycle digital waveforms are STATIC and don't give a lot of character (such as tuning drift, PWM, and other nice little oddities) that a real VCO analog waveform can do... So, sometimes for synth sounds, it's better to sample a certain length of its overall sound (i.e. the finished "filtered" sound itself, not just the waveform), and then loop it, to obtain the PWM or other character of the synth that you're sampling.

[Meatball Fulton]

It all depends on what you plan to do with the sample. Using just one sample across the whole key range sounds awful, it's what folks used to call "munchkinization" because the upper notes are simply sped up and the lower notes slowed down, only the root note sounds right. For certain sounds, this might be OK for an octave or two but if the goal is to sample real instruments it's useless.

 

Multisampling of percussive sounds is the easiest, just record the sound at as many notes and dynamic (velocity) levels as you want, record the full decay and don't loop it...done. Sampling sustained sounds are where all the heavy lifting comes into play.

 

For "one shots" or loops of drums or sound effects, multisampling isn't needed.