Weeklybeats is a 52 week long music project in which artists compose and publicly release 1 song a week for the entire year.
Starting January 1st 2024 GMT each participant will have one week to upload one finished composition. Any style of music or selection of instruments are welcomed and encouraged. Sign up or Login to get started or check our FAQ for any help or questions you may have.

WeeklyBeats.com / Music / Ipaghost's music / Personable Computer

Personable Computer

By Ipaghost on February 6, 2022 11:22 pm

Singing Sytrus! Really liked making the vocal patch, the song is reminiscent of my early 2k stuff.
Used this formant table, but I should use my own voice to make a table next time, really didn't expect it to be that easy.
Download the 91kb project here!

Such a nice guy, really

Cute-sounding vocal patch! Tight melodies too.
And what a perfect title, hehe. big_smile

Devieus wrote:

Such a nice guy, really


Just don't get a virus!

Reilly Farrell wrote:

Cute-sounding vocal patch! Tight melodies too.
And what a perfect title, hehe. big_smile


Thanks!

THE SLOWDOWN

epic.

love what you did with the voice.  I looked at that formant table and almost fainted.  I really enjoy the grindup of the drums.

orangedrink wrote:

THE SLOWDOWN

epic.

love what you did with the voice.  I looked at that formant table and almost fainted.  I really enjoy the grindup of the drums.


It's not that bad, I just used the filter frequencies for the first 3 formants of just the soprano voice and ignored everything else, but you could get away with 2. The hardest part is that Sytrus has no way of inputting numerical values for these filters so I had to tune them by hand.

heart

you hacked Luciano Pavarotti`s PC. why not.

Those vocals!! The harmonies!  Super fun.  Love the chord progressions and chippy edge too :-)  Ya tracks always put a bop in my step ;-)

Sorkfot wrote:

<3


Q-Rosh wrote:

you hacked Luciano Pavarotti`s PC. why not.


Haxored

cTrix wrote:

Those vocals!! The harmonies!  Super fun.  Love the chord progressions and chippy edge too :-)  Ya tracks always put a bop in my step ;-)


Same!

This sounds so good! For English native speakers, this is how English language music sounds to us non native English speakers. We can't really tell what is being said but enjoy the melody. Sometimes it is much better this way.

I love also the chaotic music behind, with the circus like bass around 3:20.

I need every fluid ounce of your early 2k stuff, this is super great.

Kedbreak136 wrote:

This sounds so good! For English native speakers, this is how English language music sounds to us non native English speakers. We can't really tell what is being said but enjoy the melody. Sometimes it is much better this way.

What do you mean, this is just a song in English...

Whoa, this is rad!
You program your personable computer,
Beam yourself into the reasonable future

Kedbreak136 wrote:

This sounds so good! For English native speakers, this is how English language music sounds to us non native English speakers. We can't really tell what is being said but enjoy the melody. Sometimes it is much better this way.

I love also the chaotic music behind, with the circus like bass around 3:20.


Funny, I actually think this style was directly influenced by all the French music I listened to! Jean-Jacques Perrey, Gangpol Und Mit, and DAT Politics to name a few, all have there DNA in there for sure!


ilzxc wrote:

I need every fluid ounce of your early 2k stuff, this is super great.

Kedbreak136 wrote:

This sounds so good! For English native speakers, this is how English language music sounds to us non native English speakers. We can't really tell what is being said but enjoy the melody. Sometimes it is much better this way.

What do you mean, this is just a song in English...


It might be too much for your computer! I'm not even sure where that stuff exists online anymore... might have to put it somewhere for posterity.

Himelstein wrote:

Whoa, this is rad!
You program your personable computer,
Beam yourself into the reasonable future


The problem with the future is you don't know if it will be unreasonable!

Crazy... I understand nothing in your description.  Do you have any videos of your creation process? The vocalizations are amazing and varied.  Really fun! Love the crazy chord chase progression in the second half.

NWSPR wrote:

Crazy... I understand nothing in your description.  Do you have any videos of your creation process? The vocalizations are amazing and varied.  Really fun! Love the crazy chord chase progression in the second half.



I do not have any video... I should make some but they take a lot of extra time! I can tell you how I made it though real fast.

The voice is essentially a vibrating larynx in a squishy meat tube. The larynx vibration create the pitch, or harmonics. The throat and the mouth changes shapes to create different sounds, those two independently changing cavities create two distinct resonances, which are called formants.

There are more to talking voices than just that, but that is the bare minimum we need to distinguish sustained vowels. There are actually more formants, but the relation of the 2 first formants from each other is how we can identify it as a vowel, the others give more personality (like when your nose is stuffed it sounds different). Luckily Sytrus has 3 independent filters, so it can sound slightly more intelligible.

So, what I did was create a saw wave using a sine oscillator modulating itself, which creates an FM saw wave, I also added a bit of FM vibrato with a very slow operator to make it sound more like singing. Saw waves are good because they have a lot of harmonics that can be filtered, but any wave rich in harmonics will work to varying effect (square wave sounds more robotic). Next I used the formant chart I linked above to tune each bandpass filter to match. Essentially the BP filters act as the resonating meat tubes. I could have made my own by recording my voice and noting where the peaks are on a frequency spectrum. I then repeated the process for all the formants of the 5 vowels, and set up Sytrus to change the filters in relation to it's X value. I then sum the output of the filters. So when the X controller is at 0%, it is "A", 25% E, 50% I, 75% O, 100% U. This let me change the vowel in FlStudio's piano roll very easily, and to slide between values too, which is essential for creating something that sounds like a word. I also added an envelope to add a little slide during each key press to accentuate that effect.
Since every voice is different, I simply raised all of the band pass formant values to create a "smaller" voice effect. Even if they are playing the same fundamental note, because their tubes are smaller, their formants will be higher, and their voice will seem higher/smaller.

Anywho, I essentially just ignored all the other data, like bandpass width, amount etc. To me, just having 2-3 narrow BP filters moving independently was enough to sound like a crazy singing computer! It's supper easy, you can try creating a vowel with your favorite filter!


&

emily wrote:


&


Totally excellent gif selection! Party on!

NWSPR wrote:

I understand nothing in your description.

also check this out!

https://dood.al/pinktrombone/

orangedrink wrote:
NWSPR wrote:

I understand nothing in your description.

also check this out!

https://dood.al/pinktrombone/


OMG! That is so rad!

listening again, this is really good

orangedrink wrote:

listening again, this is really good


Lol, it's pretty wacky!

Ipaghost wrote:
NWSPR wrote:

Crazy... I understand nothing in your description.  Do you have any videos of your creation process? The vocalizations are amazing and varied.  Really fun! Love the crazy chord chase progression in the second half.



I do not have any video... I should make some but they take a lot of extra time! I can tell you how I made it though real fast.

The voice is essentially a vibrating larynx in a squishy meat tube. The larynx vibration create the pitch, or harmonics. The throat and the mouth changes shapes to create different sounds, those two independently changing cavities create two distinct resonances, which are called formants.

There are more to talking voices than just that, but that is the bare minimum we need to distinguish sustained vowels. There are actually more formants, but the relation of the 2 first formants from each other is how we can identify it as a vowel, the others give more personality (like when your nose is stuffed it sounds different). Luckily Sytrus has 3 independent filters, so it can sound slightly more intelligible.

So, what I did was create a saw wave using a sine oscillator modulating itself, which creates an FM saw wave, I also added a bit of FM vibrato with a very slow operator to make it sound more like singing. Saw waves are good because they have a lot of harmonics that can be filtered, but any wave rich in harmonics will work to varying effect (square wave sounds more robotic). Next I used the formant chart I linked above to tune each bandpass filter to match. Essentially the BP filters act as the resonating meat tubes. I could have made my own by recording my voice and noting where the peaks are on a frequency spectrum. I then repeated the process for all the formants of the 5 vowels, and set up Sytrus to change the filters in relation to it's X value. I then sum the output of the filters. So when the X controller is at 0%, it is "A", 25% E, 50% I, 75% O, 100% U. This let me change the vowel in FlStudio's piano roll very easily, and to slide between values too, which is essential for creating something that sounds like a word. I also added an envelope to add a little slide during each key press to accentuate that effect.
Since every voice is different, I simply raised all of the band pass formant values to create a "smaller" voice effect. Even if they are playing the same fundamental note, because their tubes are smaller, their formants will be higher, and their voice will seem higher/smaller.

Anywho, I essentially just ignored all the other data, like bandpass width, amount etc. To me, just having 2-3 narrow BP filters moving independently was enough to sound like a crazy singing computer! It's supper easy, you can try creating a vowel with your favorite filter!

Holy smokes.. at first I thought your response was being cheeky (pun is optional) but this is a well thought out introduction to speech and how it relates to speech synthesis.  Thanks for putting this together. You could deliver a dissertation on the subject!

A lot of serious work put into this one, for sure. The amount of fluency you've reached with the voice synths is impressive.

Maybe this picture will suit you and your computing endeavours...

laguna wrote:

A lot of serious work put into this one, for sure. The amount of fluency you've reached with the voice synths is impressive.

Maybe this picture will suit you and your computing endeavours...


Thanks!

You need to login to leave a comment.
Login Sign-up