Announcement

Collapse
No announcement yet.

Production techniques

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Production techniques

    A thread dedicated to specific techniques to achieve specific sounds, e.g. drones, evolving soundscapes, otherwordly ambiences, alien noises, divine pads, abstract textures, ambient grooves and whatever else we migth be able to imagine.

    I'll start with something I'm frequently coming back to as it produces such beautiful sounds.

    Additive Resynthesis in Alchemy

    Take a field recording of e.g. a singing bird - as it will have been recorded outside there will be wind and background ambience noises in the audio which we have to remove to achieve better resynthesis results. So take a denoiser like RX 2 in denoiser mode, take a footprint of the noise you want to get rid of, overdo the denoising (in advanced mode) so that you're only left with the actual pitches the bird(s) is singing.
    Use the "Import Audio" function in Alchemy - choose additive mode and "many octaves" if the bird is singing over a wide range of pitches and chose "best frequency". After the import has taken place, turn the "P Var" knob all to the left so that all pitch modulation will turn into the overtone modulation over a fixed pitch. Turn the "Stretch Button" to the left to calm the nervous singing down. At very slow speeds this can produce beautifully evolving drones when very low pitches are played, the bird still sings but it sings in the domain of harmonics over a root note. Add some subtle pitch modulation or use external FX, to drown the birddrone in modulated/saturated/otherwordly spaces. If you're familiar with the additive editor in Alchemy manipulate the individual harmonics to sculpt the harmonic content to your taste. Then there is a plethora of modulation options inside Alchemy, you can also change the default sine wave which is used by the additive resynthesis to more complex waveforms to add rich harmonic content (in Metasynth you can even use multisamples to play back the resynthed data which is were the fun really begins).

    Also try this technique on vocal files or spoken word, where the pitch modulation of the voice turns into a spectral movement over a fixed pitch, great for robotic voices too. This can also be done in Metasynth but the procedure is a bit more convoluted, the overall sound can be better though as the resynth algos are more advanced.
    Last edited by Sampleconstruct; 12-04-2012, 02:32 PM.

    Homepage

  • #2
    And you, Sampleconstruct are the PERFECT person to start a thread like this. :cool: I can't wait to try this out. As a huge fan of Alchemy (It's my main tool in my arsenal) I look forward to gaining more knowledge about these tooks from our users like yourself.

    Cheers and keep 'em coming. :tu:
    Last edited by S1gnsOfL1fe; 05-01-2013, 11:00 PM.
    Synphaera Records
    Space | Time | Matter

    Comment


    • #3
      Originally posted by S1gnsOfL1fe View Post
      And you, Sampleconstruct are the PERFECT person to start a thread like this!!!!!!! :cool: I can't wait to try this out. As a huge fan of Alchemy (It's my main tool in my arsenal) I look forward to gaining more knowledge about these tooks from our users like yourself.

      Cheers and keep 'em coming!!!! :tu:
      Thank's - Let's see what we can come up with in this thread then...

      Homepage

      Comment


      • #4
        Here is an experiment I made today as a chillout session after too many hours of programming - some windchimes I just recorded, a one-minute long texture playing the chimes with my fingers recorded with 3 mics (L-C-R), retuned in Melodyne to an indian Raga scale and transposed downwards quite a bit, then removed the resulting artifacts and balanced the frequencies with RX2, then imported the sample into Padshop Pro using several grain streams laying out a big chord over 5 octaves in D minor (plus something), adjusting the playback speed on the fly - a tuned Highpass filter (key follow) is active in PSP, the (high) filter resonance is also tweaked on the fly so the tonality increases and decreases over time. This signal is then sent into ÜberMod inserted on the PSP track, a huge space with subtle pitch modulation, about 50% wet. Then the signal goes into a 3 band instance of Saturn, different saturation modes and feedback amounts are active in the 3 bands, some LFOs are controlling the crossover and feedback frequencies. This is then sent to a Bus where a huge dual engine B2 space is happily welcoming the signal to do it's duty.

        http://soundcloud.com/sampleconstruct/indian-chimes
        Last edited by Sampleconstruct; 05-01-2013, 11:32 AM.

        Homepage

        Comment


        • #5
          First you get me interested in Iris, then in RX2. I hope Izotope is giving you a commission. :D

          Cool experiment.

          Comment


          • #6
            Originally posted by falls a star View Post
            First you get me interested in Iris, then in RX2. I hope Izotope is giving you a commission. :D

            Cool experiment.
            Thank's - naa, no commission, just an invitation for beta-testing which I denied as and I'm already on too many beta teams it's too time consuming - I need apps that work
            RX2 is really a fantastic allrounder tool, not only for repairing/cleaning stuff but also for sample editing and sounddesign if you abuse it in a sensible way.

            Homepage

            Comment


            • #7
              I tried the demo last night. The regular version isn't that special to me (aside from a nice interface), but RX2 Advanced is really pretty amazing. Deconstruct is a wonderful thing. I can't really justify the cost right now though, sadly.

              Comment


              • #8
                Originally posted by Sampleconstruct View Post
                RX2 is really a fantastic allrounder tool, not only for repairing/cleaning stuff but also for sample editing and sounddesign if you abuse it in a sensible way.
                Could you elaborate on this? I'm always interested in subject of using and misusing tools to produce interesting sounds.
                SoundCloud // FreeSound // Twitter
                Get exposure for your electronic music through WEATNU.COM independent promotion network.
                "Shortwave" - collaboration album with Ager Sonus

                Comment


                • #9
                  Originally posted by MetaDronos View Post
                  Originally posted by Sampleconstruct View Post
                  RX2 is really a fantastic allrounder tool, not only for repairing/cleaning stuff but also for sample editing and sounddesign if you abuse it in a sensible way.
                  Could you elaborate on this? I'm always interested in subject of using and misusing tools to produce interesting sounds.
                  Sure, here are a few examples:
                  One can use the denoiser in many ways, usually you would take a footprint of the noise you want to remove and then adjust the amount of denoising to taste. But you can also filter a signal (e.g. a cello note) with a very steep bandpass, so that only a smaller frequency band is left, then use that as the "noise" footprint and apply it to the full signal. A partial of the signal will be removed, with very high reduction settings this can lead to very interesting new sounds. Or take a noise footprints from e.g. a recording in a factory and apply that to a totally different signal which has nothing to do with the original footprint source. Or take the recording of a bird flock, take a footprint of the background noise and totally overdo the reduction (100% only on the noise parameter, not on the tonality parameter) so that you're left with only the bird sounds and nothing else, sounding like under water or behind thick glass. Then import that into a sampler or granulator and start building textures, or import it into Alchemy via additive resynthesis (resynth algos hate noisy backgrounds), remove the pitch modulation, stretch it to almost standstill and you have the most beautiful spectral drone texture, as all the pitches sung by the bird are now harmonics of the root note (sorry if I'm repeating myself here).

                  Or use the spectral repair function to create new sounds. Take an audio file, select a portion in the middle of it - full frequency band, e.g. 3 seconds long, then use e.g. the "Replace" algo with enough surrounding length, so that RX2 will interpolate the portion to remove with frequencies from before and after that gap. This can yield very interesting result. Later you can then isolate only the interpolated portion of the resulting file and stretch those 3 seconds to 5 minutes with apps like paulstretch. Great for some interesting drone textures you would hardly create otherwise. And, or, or and, and, or or or or, and

                  Homepage

                  Comment


                  • #10
                    A piano impro played simultaneously by several students of musicology who participated in a seminar about electronic sound generation which I conducted at the university in Münster/Germany.
                    This impro was processed with 2 convolution reverbs, each one using a segment from another impro we did on the same day as an impulse response. Some algorithmic reverb (B2) and EQ was added in the mix, a modulated stereo spreader is active on one of the convolution reverbs. No dry signal is audible throughout the track, it's all the convoluted signals.

                    http://soundcloud.com/sampleconstruct/convoluted-mystery-seminar

                    Homepage

                    Comment


                    • #11
                      Another convolution experiment:
                      Experiment processing a physically modelled bell sound
                      (playing a slow tonal impro) with a multiband convolution plugin (Melda). Each of the 3 bands carries a different female vocal sample (recorded last week for a sounddesign project), the band crossover frequencies are modulated by a slow LFO. Some VVVerb was added in the Mix.

                      https://soundcloud.com/sampleconstruct/convoluted-vox
                      Last edited by Sampleconstruct; 07-28-2013, 01:52 AM.

                      Homepage

                      Comment


                      • #12
                        Originally posted by Sampleconstruct View Post
                        Another convolution experiment:
                        Experiment processing a physically modelled bell sound
                        (playing a slow tonal impro) with a multiband convolution plugin (Melda). Each of the 3 bands carries a different female vocal sample (recorded last week for a sounddesign project), the band crossover frequencies are modulated by a slow LFO. Some VVVerb was added in the Mix.

                        https://soundcloud.com/sampleconstruct/convoluted-vox
                        Beautiful!
                        http://silentfrill.bandcamp.com/

                        https://manducator.bandcamp.com/

                        Comment


                        • #13
                          Originally posted by Sampleconstruct View Post
                          Another convolution experiment:
                          Experiment processing a physically modelled bell sound
                          (playing a slow tonal impro) with a multiband convolution plugin (Melda). Each of the 3 bands carries a different female vocal sample (recorded last week for a sounddesign project), the band crossover frequencies are modulated by a slow LFO. Some VVVerb was added in the Mix.
                          Wow that's really nice. I'm soooooooo excited about writing my own convolution software through SuperCollider --- now that your example is proof (to me) that audio convolution can be so interesting. As an astronomer and programmer, my main line of work was high resolution astronomical imaging. So all of the optical "tricks" that I've learned over the past 20+ years in astronomy can pretty much be directly applied to audio. Waves is waves as far as the fundamental algorithms go. The "bible" of optics and a very much used book in my library is Max Born and Emil Wolf's 'Priciples of Optics'. Most of what's contained in this book can be directly applied to audio. Is there an equivalent bible for audio?

                          I just gotta get over this stupid hurdle of learning yet another f*cking programming language! Argh!!!!!

                          Peace to All

                          Comment


                          • #14
                            Originally posted by AstroAndMusic View Post
                            Originally posted by Sampleconstruct View Post
                            Another convolution experiment:
                            Experiment processing a physically modelled bell sound
                            (playing a slow tonal impro) with a multiband convolution plugin (Melda). Each of the 3 bands carries a different female vocal sample (recorded last week for a sounddesign project), the band crossover frequencies are modulated by a slow LFO. Some VVVerb was added in the Mix.
                            Wow that's really nice. I'm soooooooo excited about writing my own convolution software through SuperCollider --- now that your example is proof (to me) that audio convolution can be so interesting. As an astronomer and programmer, my main line of work was high resolution astronomical imaging. So all of the optical "tricks" that I've learned over the past 20+ years in astronomy can pretty much be directly applied to audio. Waves is waves as far as the fundamental algorithms go. The "bible" of optics and a very much used book in my library is Max Born and Emil Wolf's 'Priciples of Optics'. Most of what's contained in this book can be directly applied to audio. Is there an equivalent bible for audio?

                            I just gotta get over this stupid hurdle of learning yet another f*cking programming language! Argh!!!!!

                            Peace to All
                            That sounds intriguing, how can we help you to get over that hurdle and start programming NOW

                            Homepage

                            Comment


                            • #15
                              Originally posted by manducator View Post
                              Originally posted by Sampleconstruct View Post
                              Another convolution experiment:
                              Experiment processing a physically modelled bell sound
                              (playing a slow tonal impro) with a multiband convolution plugin (Melda). Each of the 3 bands carries a different female vocal sample (recorded last week for a sounddesign project), the band crossover frequencies are modulated by a slow LFO. Some VVVerb was added in the Mix.

                              https://soundcloud.com/sampleconstruct/convoluted-vox
                              Beautiful!
                              Thank you

                              Homepage

                              Comment

                              Working...
                              X