Advice on how to lay out key ranges to samples

• WF 9 0

  Message from Wanda Fish on …
  Soundfonts ideally map each key to an individual sample.
  Otherwise (e.g. for size/efficiency/laziness) whole key ranges are
  mapped to one sample. Non-root keys will be "interpolated" so to speak.
  
  In case I do still have the opportunity - how should I 'lay out' my samples to key ranges?
  Let's say I'm sitting here with a strange rare instrument having one hour studio time.
• Message from Wanda Fish on … 1
  Probably someone is going to reply "if it sounds good, you are good"
  but I'm interested in general advice here - and you just get one try with a Stradivari
  
  Just to detail my (made up) use case:
  
  Let's say the instrument can produce pitches in 3 ocatves, e.g: c2 to c5.
  My gut feeling is: sampling just six notes should be good enough?.
  I'd then go for the extreme ones c2 and c5, plus maybe c3 g3 c4 and g4.
  Or not?
  
  Il also I wonder if my ranges should "favour" downsampling where possible, instead of
  having root keys in the middle of a range (as many downsampled keys as upsampled ones)
  
  PS: Polyphone is a great piece of software!
• 70 0

  Message from Sylvia on … 1
  The scenario that you've described is 'sampling in fifths'. If you only had one hour with an instrument, I would recommend you to sample in thirds or per key.
  
  You can sample in many ways:
  
  In octaves - Low quality, most compact
  In fifths (C, F# or G) - Medium quality, fairly compact
  In thirds (C, Eb, F#, A) - High quality, fairly large
  Per key (All keys) - Highest quality, largest
  
  You should select the one that you want.
  
  When you sample acoustic instruments, don't favour downsampling. You would not get the most natural sound, as if you have samples in thirds, putting root keys in the middle will result in interpolation of no more than one semitone, resulting in a better sound. However, for synthetic sounds, downsampling has less of a negative effect on sound quality.
• WF 9 0

  Message from Wanda Fish on …
  I'd better put the actual root key in the middle of the corresponding range.
  Convinced.
  
  Btw I only now re-realize how little I know about filter design, Nyquist frequencies and whatnot
  - but that's another story altogether.
• 417 0

  Message from Davy on … 1
  I extensively used Polyphone for creating organs based on real samples and to my experience, I prefer downsampling because some high harmonics sometimes scream. Let's say that when I am using a sample, I extend the range to the bass 50% more than I would do in the trebles.
  
  Regarding the number of sample you should use, it depends on the harmonics, the evolution of the timbre across the keyboard, the stability of the sound. The more harmonics, evolution and instability => the more samples you will need.
  
  With the example of the organ, I had no problems with flutes but more with the trumpets. In the latter case I had to select samples that "go together", having a similar attack, a similar harmonic content, etc.
  
  Your hears will do the job: first place all the samples over the keyboard, try to correct the attacks, the attenuation, and test it in using every key from the first C to the last one. This could be a hard job but always interesting
• WF 9 0

  Message from Wanda Fish on …
  The "high notes and rich harmonics samples tend to scream when being upsampled" obviously point to the Nyquist frequency problem
  I guess I'm going to favour downsampling at least for the very high registers (i.e. record the c5 in my example) and
  use the middle-of-the-range approach elsewhere.

Sign in or register to take part in discussions.