Adding Bow Noise

Tutorial home

Drawing a bow across a string creates a scratchy noise that makes a noticeable addition to the sound of the instrument. It’s perhaps analogous to the breath noise in a blown instrument.

We’d like to model this, and there are a couple of different approaches we can use. For example, we could mimic what we did with blown pipes, and add noise to the bow velocity input.

But there’s a better way. The pulsed noise method that we used in some of the earlier models was originally developed with bowed stringed instruments in mind. The developer of this technique, Dr. Chris Chafe, is in fact a cellist.

How we’ll do it

This technique is based on the observation that noise is added to a bowed instrument when the string is slipping under the bow. When the string is stuck to the bow, no noise is created.

We’ll follow these steps, which are similar to what we did with the reed woodwind model:

• Create sparse noise.
• Place a level detector before the bow table, and use that to create pulses which are used to turn the sparse noise on and off. The level detector’s function is to determine when the bow is slipping. In our particular Friedlander-Keller function, the bow begins slipping when the input to the function is greater than 16 (or less than -16). Listening tests will cause us to back off our level detector a little, to 20.
• Include a lowpass filter to provide some basic tone color.
• Inject this noise back into the model after the bow table.

Four parameters control the noise, and are brought out to the front panel:

1. Noise density (the sparseness of the noise).
2. Trigger threshold (which determines the width of the pulses).
3. Lowpass filter cutoff frequency.
4. Output level.

A patch is below.