posted 26 March 2014 15:13            

I have a sequence of drums I have played and i did some sound design to them but still they sound acoustic and im looking to spice up the timbre of them, i dont really want to use an analogue drum machine instead, there are something about acoustic ones i like (formants)

so i thought that maybe i could do some convolution or morphing:

what i thought was to have the sequence of the drums in one file and a wide spectrum timbre I have created with synths etc in another file so my goal is to blend that synthetic timbre with the one of the drums but i still want to be able to hear the drum sequence and not just a mass of sound, i want to hear the drums with the sequence and all that just with a bit of a more synthetic sound, i don't want to lose the performance in the sequence.

any suggestions as what kind of prototype i could use to achieve that?

posted 28 March 2014 22:13            

I like to take a bar or two of drums, edit some form of vocal or speech material so that it hits with the main transients, export them so they are both the same exact length.

I usually try a few different permutations of the following.

Use the crossfilter prototype to produce a convolution, gain is everything. (nasty clicking if too loud)

Create a few of the different analysis files for each type, also try putting both in the TAU, saving and then there are some interesting starting points in the Galleries. One of the resynthesis types is based on noise, I forget which, this can be great for speech and drum combinations

The main secret to success is trying to match up as much of timing of the material as possible beforehand. Although it is quite fun to do this in TAU, but you really need a solid starting point.
I usually end up submixing the results back in with the original drums, or using it for edit moments.

Splendid.

posted 29 March 2014 14:34            

Just to add a little note about crossfilter clipping.

The thing about the cross filter is that the output level can vary vastly depending on the length of the response and the similarity of the content of the two signals. Very few modules have such a hugh range of potentially different output levels and getting the level right would be virtually impossible if you didn't have the built in level control of the cross filter.

It is very important to use the level control that is built into the crossfilter to get your crossfilter levels right. Unlike other level controls, this level control will accept values greater then one if that's what you need to get the level right. Reducing the level before the input means that you are not making optimum use of the dynamic resolution, and adjusting the gain after the output is like closing the door after the horse has already bolted, and will not avoid clipping.

The built in level control, adjusts the gain just after the point where the two signals are merged, but the resulting signal is still in an ultra high resolution/high headroom state.

The rule is that if the inputs are not clipping, then the the crossfilter will not clip, as long as you have set the crossfilters built in level parameter correctly, such that the output is not hitting +/-1. If you try to attenuate anywhere else, then you are not getting the optimum out of the crossfilter.

posted 30 March 2014 20:16            

Thanks for the tips!!

Charlie

posted 01 April 2014 18:11            

quote:

---

Originally posted by SSC:
One approach might be to create a CrossFade between the original drum on one side. And, on the other side, the new timbre multiplied by the amplitude envelope of the drum. For example, you could use a Product to do the multiplying by either a PeakTracker or AmplitudeFollower on the original drum to extract its amplitude envelope.

Another approach is to use the Vocoder with your original drum as the SideChain and the new timbre as the Input. Then mix the Vocoder with the original signal so you can get any combination of the two sources.

---

posted 01 April 2014 21:26            

quote:

---

Originally posted by pete:
Hi Folks

Just to add a little note about crossfilter clipping.

The thing about the cross filter is that the output level can vary vastly depending on the length of the response and the similarity of the content of the two signals. Very few modules have such a hugh range of potentially different output levels and getting the level right would be virtually impossible if you didn't have the built in level control of the cross filter.

It is very important to use the level control that is built into the crossfilter to get your crossfilter levels right. Unlike other level controls, this level control will accept values greater then one if that's what you need to get the level right. Reducing the level before the input means that you are not making optimum use of the dynamic resolution, and adjusting the gain after the output is like closing the door after the horse has already bolted, and will not avoid clipping.

The built in level control, adjusts the gain just after the point where the two signals are merged, but the resulting signal is still in an ultra high resolution/high headroom state.

The rule is that if the inputs are not clipping, then the the crossfilter will not clip, as long as you have set the crossfilters built in level parameter correctly, such that the output is not hitting +/-1. If you try to attenuate anywhere else, then you are not getting the optimum out of the crossfilter.

I hope this makes sense.

---

posted 01 April 2014 21:28            

quote:

---

Originally posted by CharlieNorton:
Hello!

I like to take a bar or two of drums, edit some form of vocal or speech material so that it hits with the main transients, export them so they are both the same exact length.

I usually try a few different permutations of the following.

Use the crossfilter prototype to produce a convolution, gain is everything. (nasty clicking if too loud)

Create a few of the different analysis files for each type, also try putting both in the TAU, saving and then there are some interesting starting points in the Galleries. One of the resynthesis types is based on noise, I forget which, this can be great for speech and drum combinations

The main secret to success is trying to match up as much of timing of the material as possible beforehand. Although it is quite fun to do this in TAU, but you really need a solid starting point.
I usually end up submixing the results back in with the original drums, or using it for edit moments.

Splendid.

Charlie

---

I originally was thinking to try it on tau but was not sure, thanks for the post, I will try it now, looks like a lot of fun!!!!