how do I do this?

iskra · Post by **iskra** » Thu Oct 05, 2006 7:49 pm

a few questions really...

I have a 15 second spoken sample that I want to map to the pitch of the melody/riff of my song. I have STS5000. The main riff is recorded as midi, I want to use this to trigger the spoken sample, but for the note data not to retrigger the sample, but for the sample to play in its entirety & for the note data to change the pitch of the sample to match the melody.

Second question, I have a few samples recorded from a documentary, which have low level music in the background, I have tried to remove this music with eq with imperfect results (I can use the phase trick to remove the vocals perfectly, but that's the opposite of what I want). I realise that perfect results might not be possible, but is there a better method? I think there are programs where you can sample a noisy background to remove it (this might work where the music is a low drone), can anyone recommend any, & how well do they work?

Counterparts · Post by **Counterparts** » Fri Oct 06, 2006 1:53 am

Hi there

To change the pitch of your sample as you indicate, I can think of three possibilities.

The first would be to generate suitable pitch bend messages based on the MIDI note of your sequence, but this would be difficult to get right; you'd need some kind of algorithm or look-up table operating to convert the MIDI note number (0-127) into the correct pitch bend value.

The second would be to use a wave editor to timestretch the various sections of the sample (as defined by where the MIDI notes occur) so that the pitch of the sample is changed at various points to match the MIDI notes.

The last way I can think of would be to use a vocoder; use the MIDI notes to produce a sound to feed into the vocoder as the control signal, effecting the vocal sample in the way that you want. You'll need to select an appropriate signal and vocoder settings such that it doesn't colour the original sample too much, but I think this would be the easiest method to use.

As for removing the background music, unless you have this recorded as seperately, this would be very difficult to acheive. You could perhaps try running it through an expander, then a low noise or rumble filter, then compress the signal again. I doubt you'd get perfect results though.

HTH!

alfonso · Post by **alfonso** » Fri Oct 06, 2006 1:54 am

On 2006-10-05 20:49, iskra wrote:
a few questions really...

I have a 15 second spoken sample that I want to map to the pitch of the melody/riff of my song. I have STS5000. The main riff is recorded as midi, I want to use this to trigger the spoken sample, but for the note data not to retrigger the sample, but for the sample to play in its entirety & for the note data to change the pitch of the sample to match the melody.

Second question, I have a few samples recorded from a documentary, which have low level music in the background, I have tried to remove this music with eq with imperfect results (I can use the phase trick to remove the vocals perfectly, but that's the opposite of what I want). I realise that perfect results might not be possible, but is there a better method? I think there are programs where you can sample a noisy background to remove it (this might work where the music is a low drone), can anyone recommend any, & how well do they work?

For the sts 5000 thing there is a setting that can make you add an harmonization at a certain point of a sample, but I never tried to just "continue" a single voice, maybe it is possible just stopping the first note immediatly after the second has started...I don't know if pitch wheel is time compensated as well, never tried, but it could be, you could program discrete values in your sequencer..

For the secont issue you could try:

minimize the music presence with eq---> A

apply to A the phase trick and remove vocals---> B

Reverse the phase of B and mix with A, you should have removed a lot of the remaining music in A ----> C

Use a noise gate on C with very high threeshold and pretty short release...----> D = definitive.

This is what can I imagine to get the maximum possible results....

Counterparts · Post by **Counterparts** » Fri Oct 06, 2006 1:56 am

alfonso wrote:
...you could program discrete values in your sequencer..

Now that's a good idea

alfonso · Post by **alfonso** » Fri Oct 06, 2006 4:00 am

I was reading on the manuals downloaded from CWA site (I'm not in the studio) that you can enable the "Mono Legato" function in the Program Drawer under the main sts5000 panel that is used to avoid retriggering, so you can keep the previous note pressed but only get the sound from the new one. It's related to "Robot" mode (all the phrase playing in a single pitch determined by the key), but it might work also in "Autochord" mode (the phrase mantains it's own pitch modulation and it's all transposed with the new notes, from the point in wich you add them, summed to the previous ones).

Try...

iskra · Post by **iskra** » Fri Oct 06, 2006 4:53 pm

thanks counterpoint & alfonso for all suggestions

I had thought of a few already that have been mentioned, I thought of using a wave editor to draw in pitch changes, but I've never used midi with a wave editor & dont know if its possible to trigger pitch changes from a midi sequence. If it is this would be good, if not trying to match up the timing of a midi sequence would be extremely tedious.

I also tried vocoder method with not very good results, but admittedly I didnt play with it for very long, maybe I will try again.

alfonso, I am not quite sure what you mean by programming discrete values in my sequencer? values of what?.....there is a logical editor that will allow me to turn the note data into other midi data which might be useful & I have found a midi transformer device on planetz (created by you

) that might be useful

I will also play with mono legato & autochord functions in STS

you have given me a lot to think about, I will report back if I have success with any of these methods

finally

apply to A the phase trick and remove vocals---> B

Reverse the phase of B and mix with A, you should have removed a lot of the remaining music in A ----> C

I tried this without the other steps (eq first & noise gate after) & it didnt really work, because after removing the vocals you end up with two mono signals & reversing phase of this back in with original doesnt really work (it makes a very small difference), but I will try the other steps & see if that makes any difference

alfonso · Post by **alfonso** » Fri Oct 06, 2006 5:11 pm

For discrete values I mean to input in sequencer directly the pitch bend values that can change the pitch of the note sent.
.........
It was not referred to Wave editor, but to sts 5000 that has formant and time correction capability. The manual is very important to read!

jea · Post by **jea** » Sat Oct 07, 2006 7:42 am

Hi,

try some M/S technique, it might work.

Or any combination.

Best wishes,

iskra · Post by **iskra** » Sat Oct 07, 2006 9:53 pm

I was reading on the manuals downloaded from CWA site (I'm not in the studio) that you can enable the "Mono Legato" function in the Program Drawer under the main sts5000 panel that is used to avoid retriggering, so you can keep the previous note pressed but only get the sound from the new one.

thanks alfonso, the mono/legato switch was exactly what I was looking for

now I'll go back & work on the music removal problem