How many velocity/volume levels are needed to create a quality sound?
I am thinking about sampling some sounds from a concertina. I can see how to capture the minimal and maximum volume levels, but what about the intermediate levels?
I could capture a few different volume levels, but how to assign a velocity to such samples?
Please check the following link blog post for a picture and explaination.
Sorry about the double, editing is not possible apperentally.
please also check the FAQ question number #6
Thanks, I read that but it does not really answer my question.
@Netanel: yes, editing not possible yet ;)
@Don: it's a difficult question. It really depends on the instruments:
for a church organ, as it's not velocity sensitive, 1 velocity layer is enough to provide a good quality
for a piano, something like 4 or 5 velocity layers is a minimum to reproduce the differences of feeling; idem for a Rhodes electric piano
How to build such a sample-set? It's simple. Record at 4 different volumes and name the files like this:
C1_vel32.wav, C1_vel64.wav, C1_vel96.wav, C1_vel127.wav
C#1_vel32.wav, C#1_vel64.wav, C#1_vel96.wav, C#1_vel127.wav
...
Then create a definition.txt file with a single line %notename_vel%velocity.wav and that's it!
Good luck!
Jo
Thanks for this.
If I do have 4 or 5 different levels then how can I objectively assign a level value to the samples?
Does it make sense to assign it based upon the peak decibel level?
If so, say I have 4 samples at 10, 27, 42 and 62 db, for example, then can I map the db values linearly to velocity levels in the range 1 to 127?
Thx.
Don.
Yes I think it makes sense to assign it based upon the peak dB level, or I would probably go for the average sound level dB "RMS". (Sony Soundforge computes it well).
Both would make sense.
About how to map them to velocity in the range 1-127, I would first try linearly, then if I don't like it, I would modify a bit according to my tastes.
I don't have a real rule for that, I would go for trial and error.
Sorry Don, I thought you were asking how to sample an instrument. I misunderstood.
Try to think about it this way,
there's a limitless amount of velocities in which you can play an instrument such as the concertina. This is impossible to translate accurately into a 1-127 range. it simply would never be enough.
The classical way of thinking would be to record the velocities in dynamics, that is:
Very, very low velocity: ppp = Pianississimo; Very, very, softly.
Very low velocity: pp = Pianissimo; Very softly.
Low velocity: p = Piano; Soft.
Moderately low: mp = Mezzo-Piano; Moderately soft.
Higher than average: mf = Mezzo-Forte; Moderately loud.
High velocity: f = Forte; Loud.
Very high velocity: ff = Fortissimo; Very loud.
Very, very high velocity: fff = fortississimo; Very, very loud.
That being said, more often than not this is NOT what you would find in an average sampler.
Analyzing professional instruments such as the Vienna Symphonic Library, most instrument has 4 velocity layers, 5 maximum. the Roland RD 700 for example, uses 4 velocity layers.
In classical thinking that would be:
Low velocity: p = Piano; Soft.
Moderately low: mp = Mezzo-Piano; Moderately soft.
Higher than average: mf = Mezzo-Forte; Moderately loud.
High velocity: f = Forte; Loud.
So by using 4-5 velocity levels, you're on the right track.
This thread talks about the transition between velocity and dB. Maybe you'll find it useful. It's not exactly the same as it uses a different sampler, but it's generally the same thing.
If you dig around, you'll find that every sampler does it differentally. So my advice would be - Trust your ears!
Thanks for all of this useful advice.