oVaVoX
Experimental "singing" monophonic / polyphonic analog additive synthesizer with two oscillators containing tonal and noise sawtooth curves, with 3 parallel bandpass filters, with 2 LF Oscillators for loudness and phase modulation. Has fine tuning of oscillators, oscilloscopes and peak meters. Includes a joystick for the tone wheel and cutoff of all three filters.
In parallel with the creation of various virtual synthesizers, I work in the field of creating "virtual vocalists", recording quite natural-sounding vocal parts for some of my own musical recordings. I’m exploring various ways of simulating a human voice on a computer, in particular, recording realistic singing in Russian, because I am a misanthropic and asocial creative artist who does not want to spend time inviting, maintaining and recording live singers. And I already have some successes, although, of course, the quality of my vocal recordings still leaves much to be desired, ah-ah ...
Empirically, I was able to establish that the highest quality results on simulating human singing on a computer can be obtained using the following programs:
OvAvOx was created easily and pleasantly: first, I remade the qUADRovECTRO interface to fit the color of the human body, then I added 2 LFOs synchronized with the tempo, tuned 4 analog oscillators in a special way and passed them through 2 Moog (c) low-pass filters. And a miracle happened: Synthedit sang. At first he sang like a boy and a girl. Then he sang, like an opera diva and a jazz singer. And then he sang, like ... an opera soloist and an unhappy man, heh, and this male voice reminded me very much of the Singing Synth patch for the Subtractor synthesizer in Reason Factory Sound Bank.
Further - more: I decided to connect a Vowel filter at the output, because Moog (s) filters did not give out the naturalness of the voice I needed. What started here, my previously created children, women and men sang as if they sounded from an old vinyl record, and even as lively as if they were real live singers. As a result, I debugged the overall structure a bit, made the interface more visible, and added a few more patches with voices. And among them: a voice uttering "WoW"; voice pronouncing "Oy"; voices with breath, voices with the sound of a piano - and even patches with imitation of hammer percussion instruments.
The structure of the device and the principle of synthesis:
4 oscillators generate sine, saw, ramp, triangle, white noise and pink noise, LFO MIX is responsible for their subtractivity, which does this on the basis of divisions with reference to the tempo of the sequencer or host. Please note that it works the other way around: turning the amount knob to the right decreases the force, and turning it to the left increases it.
LFO PWM generates low-frequency pulses for parameters of all 4 oscillators at once, such as Pulse Width and Modulation Depth. It is he who creates the very vibrato, which is an important component of any voice. Please note that the rotation of the amount control to the left of the center produces clicks (unfortunately, this is a design flaw, and I still do not know how to fix it). Therefore, I do not recommend this position of this handle to you.
2 Moog (c) low-pass filters are 1-stage voice filters: they set the initial resonance and cutoff level, which then passes through the Vowel 2-stage filter, simulating a certain, fairly accurate, similarity to human singing and speaking voices. Of course, this is just an imitation, but how good it is, IMHO.
There are frequency analyzers for each oscillator and filter, as well as 8 bulbs in the lower and upper part of the instrument that light up every time you press on its keyboard or on the keyboard of an external MIDI controller.
Empirically, I was able to establish that the highest quality results on simulating human singing on a computer can be obtained using the following programs:
- Vocaloid (in part, since this program has a lot of sound banks with a non-natural sound) (in addition, it is intended primarily to simulate singing in Japanese and Latin, but not for Russian, oh-oh)
- TTS with samples of Russian speakers, extended over time in Ableton Live Warp and passed through AutoTune (this method allows you to almost perfectly imitate Russian-language singing vocals, however, due to the abundance of all kinds of speech artifacts when playing many sound banks, this method is also not very good, In addition, copyright holders for TTS banks do not sell licenses for their use in show business, uh-huh)
OvAvOx was created easily and pleasantly: first, I remade the qUADRovECTRO interface to fit the color of the human body, then I added 2 LFOs synchronized with the tempo, tuned 4 analog oscillators in a special way and passed them through 2 Moog (c) low-pass filters. And a miracle happened: Synthedit sang. At first he sang like a boy and a girl. Then he sang, like an opera diva and a jazz singer. And then he sang, like ... an opera soloist and an unhappy man, heh, and this male voice reminded me very much of the Singing Synth patch for the Subtractor synthesizer in Reason Factory Sound Bank.
Further - more: I decided to connect a Vowel filter at the output, because Moog (s) filters did not give out the naturalness of the voice I needed. What started here, my previously created children, women and men sang as if they sounded from an old vinyl record, and even as lively as if they were real live singers. As a result, I debugged the overall structure a bit, made the interface more visible, and added a few more patches with voices. And among them: a voice uttering "WoW"; voice pronouncing "Oy"; voices with breath, voices with the sound of a piano - and even patches with imitation of hammer percussion instruments.
The structure of the device and the principle of synthesis:
4 oscillators generate sine, saw, ramp, triangle, white noise and pink noise, LFO MIX is responsible for their subtractivity, which does this on the basis of divisions with reference to the tempo of the sequencer or host. Please note that it works the other way around: turning the amount knob to the right decreases the force, and turning it to the left increases it.
LFO PWM generates low-frequency pulses for parameters of all 4 oscillators at once, such as Pulse Width and Modulation Depth. It is he who creates the very vibrato, which is an important component of any voice. Please note that the rotation of the amount control to the left of the center produces clicks (unfortunately, this is a design flaw, and I still do not know how to fix it). Therefore, I do not recommend this position of this handle to you.
2 Moog (c) low-pass filters are 1-stage voice filters: they set the initial resonance and cutoff level, which then passes through the Vowel 2-stage filter, simulating a certain, fairly accurate, similarity to human singing and speaking voices. Of course, this is just an imitation, but how good it is, IMHO.
There are frequency analyzers for each oscillator and filter, as well as 8 bulbs in the lower and upper part of the instrument that light up every time you press on its keyboard or on the keyboard of an external MIDI controller.
My sincere thanks to Jeff McClintock for the Cinestar SynthEdit 1.1850 Unicode VSTi programming environment. (Copyright 2000-2009 Jeff McClintock, https://www.synthedit.com/)
Free Download:
Normal working in Windows XP, Windows 7, Windows 8 & Windows 10 Pro x64
Long-awaited update!!!
(Starts and runs normally on Microsoft Windows 10 Pro x64,
not tested on other operating systems)
The archive is password protected!
Password:
ovdielengineering.blogspot.com
(Enter without spaces, hyphens, underscores, quotes and brackets, completely and completely, including all periods !!!)
Attention!
- During its launch in host or sequencer, this synthesizer extracts files with the extension * .SEP and * .SEM. These files are NOT malware, but are separate synthesizer modules. They DO NOT harm your system. Scanned ESET Smart Security 14, no malware found.
- This synthesizer is capable of producing a high level of bandpass filter resonance as well as producing loud, overdriven sounds! Be careful with knobs that change resonance and volume, they can damage your hearing or negatively affect your mental state!
- This software synthesizer is made in the "SynthEdit" modular constructor by Jeff McClintock. This development environment is widely known for producing VST synthesizers with an increased load on the central processor! Alas and oh, but so far I have not figured out how to fix this, so strictly monitor the number of instances of this synthesizer in your DAW!
Donate!
If you are satisfied with this work and want to reward its author for the work done, then you can donate any amount of money to pay for its further development. If it does not bother you, transfer USD, RUB, BTC or ETH to the purses indicated below:
PayPal:
QiwiWallet:
BitCoin:
17fUsW8jEuXQTAFLxH1FvRUhTg3NomYUKi
Ethereum:
0x6bfF375A5422c249fbf832303BE7690a20500Ecc
