• Hi let's make some vocoder effect
  • This is ems vocoder 2000 I want to make it
    1600 x 900 - 138K
  • Here's a good video rundown of how it works - anyone willing to help on this?

  • It looks like it does some magic by letting through the initial noise burst (voiced consonants) to make it a little more intelligible.
  • The voiced/unvoiced detector is a high pass and low pass filter each fed into an envelope follower. The output of the envelope followers is then compared. If the low pass is higher than the high pass, the sound is considered voiced, otherwise it's unvoiced. I had a pretty good link but I can't seem to find it now. The EMS 2000. was a 16 band vocoder, so you would need 16 filters plus the ones for the voiced/unvoiced detector. According to one description I found the filters were 6th order devices.
  • Didn't @Dcramer make a vocoder in Audulus?
  • http://forum.audulus.com/discussion/20/new-patches/p3

    Yeah he did! Unfortunately it is ancient... BTW did there used to be a sample node?
  • I found the link I looked at before. This is an example patch using the Nord Modular soft-synth. It uses 8 bands with 2nd order (12dB/octave) filters. It should port to Audulus pretty easily. The original EMS configuration of 16 bands of 6th order filters might be too much CPU load. I forgot you need a pair of filters for each channel so that would make 34 filters in total. Here's the link: http://electro-music.com/nm_classic/014_threads/Folder/Speech/SpeechSynthesis&Processing.htm
  • Just as an experiment I built a 16 band unit with 6th order filters and as I expected, I'm at 100% on my iMac even without the voiced/unvoiced detector pulse generator etc. I'll try less bands and lower order filters.
  • I actually use the Nord vocoder! I have a micromodular from back in the day and the vocoder is one of the presets that is left on it. I can vouch for 8 bands of 2nd order filter.
  • It looks like 8 bands using 2nd order filters will be OK from a CPU load perspective, at least on the Mac. I’m having some problems with the mic and ADC nodes on my Mac so I don’t currently have any way to get an audio signal into Audulus. Testing will have to wait until I get it sorted out.
  • Made one eons ago (5 years actually! Wow)


    Not sure if this still opens in recent versions, will try dig it up
  • Nice! Audulus 2 style :D
  • Yes but it's so bad that adulus 3
    can't load projects which are made with adulus 2 version
  • Vocoder has an internal oscillator so you can talk without playing synthesizer . So how to make it?
  • The 16 BP filters and an envelope follower is not a problem
  • How to make for example left is a modulator and the right is a carrier?
  • I've switched to bi-quad based filters from the SVF model I was using earlier and it's a much lower CPU load so I was able to raise the number of bands back to 16 with 6th order filters and still run it on my iPad Air 2. This is still a work in progress but this is what I've got so far.You can use an external source as carrier or an internal oscillator. The internal oscillator has level and frequency controls and wave type. There is a toggle to switch carrier sources. I have also included a voiced/unvoiced detector with a toggle to switch it in and out. There is a level control for the noise level. There are also controls for vocal level and smoothing which controls the smoothing on the envelope followers. It still needs a few tweaks and UI polish but I think it sounds pretty good.

    9:30 PM - updated UI - new file
    Screen Shot 2018-01-23 at 6.40.14 PM.png
    2452 x 1644 - 321K
    16 band vocoder biquad.audulus
  • So you had made voiced/unvoiced module?
  • @J12345 yeah - says so just above, it comes with a toggle switch to turn it on and off. Right in @stschoen!!
  • It's interesting
  • But how now use it if I have a lots of vsts such as moog arp prophet etc. So I want their sounds to be vocodered
  • I use ableton but I don't like their vocoder. I want to make voice like daft punk
  • To vocode an external audio source (rather than the internal oscillator) switch the external toggle on (red) and feed the source (Moog, ARP etc) into the ext input. Feed your vocal signal into the vocal input and the vocoded source will be on the output. The voiced/unvoiced detector is currently configured so that it’s only used with the internal oscillator although that could easily be changed. I figured if you were using an external carrier you probably wouldn’t want noise inserted.
  • How you did modules you programming?
  • @J12345 - Based on what you just said and other questions you had before, I’m not sure if you understand that Audulus modules are made with Audulus nodes, and not in some other program like a C++ compiler. Pardon me if I’m wrong thats just the sense I’m getting - but Audulus is visual programming, so you have these little packets of premade code that you string together. So stschoen programmed this module in Audulus. I suggest checking out the Modular Synthesis 101 tutorial pinned to the forum front page to watch how a module comes together. Again, sorry if I’m wrong, but I think you asked elsewhere how Audulus modules are coded. They *are* packets of C++ code, but that’s the side of Audulus only Taylor, the developer, sees.
  • I understand
  • I updated the UI slightly, increased the maximum gain on the external carrier and vocal inputs, added a wet/dry mix knob and changed the voiced/unvoiced detector so that it works on either an internal or external carrier. Since it can be switched on and off separately I thought that made more sense. I also added audio indicators on the external inputs. They actually measure the level post-gain but the input port circles seemed the best place to put them. Blue for very low or no signal, green for audio, yellow for >95% and red for clip. They're logarithmic response and damped. A bit like the level rings on my Focusrite.
    16 band vocoder V1.1.audulus
    Screen Shot 2018-01-25 at 4.42.25 PM.png
    1522 x 1456 - 212K