Synthesizer V - New voice synthesizer for Windows & Linux (Mac support added)

VST, AU, AAX, etc. plug-in Virtual Instruments discussion
ENV1
KVRAF
2731 posts since 31 Aug, 2011

Post Thu Aug 30, 2018 7:48 am

wagtunes wrote:Not sure what you did but the vocals sound human. Could use some boosting in volume (the vocals) but otherwise, well done.

I'm impressed.

Thanks.

Like i said, for 1984 technology its certainly OK.


As to what i did, nothing really. I just picked a voice preset that sounded halfways acceptable at the required pitches and speeds (Hyper Male in this case) and started sampling away. Other than cutting some bits out* its all exactly as it came out of TalkIt, plus delay and reverb of course to mask the rough stuff a bit.

(*The Ds for instance always have that extra bit at the end that sounds like 'duh', so if you wrote 'and' it would sound like 'and-uh', so i simply cut the pesky 'uhs' away. And there were a few others with similar bits that had to go.)


As to loudness, i just noticed that BuzzMaxi actually made the vox a lot louder. Because of this you can now hear some artificiality bits that i cant hear without the level boost even when i raise the speaker volume. (No big deal though since im not through with the levels anyhow.) But yeah, they are kept quiet on purpose, the FX help only so much, the rest is keeping the level low so the bits where it sounds very synthetic dont stick out like sore thumbs.

User avatar
vurt
addled muppet weed
68538 posts since 26 Jan, 2003 from through the looking glass

Post Thu Aug 30, 2018 10:09 am

given the additional helixbeta challenge, that is even more impressive!
but even if I had just heard this, with out any knowledge of the work that's gone in to it, its still a damn fine piece of work!
hearing those vocals, you would never imagine its some old piece of software.
thanks for sharing, and all that work you did, well worth it!

ENV1
KVRAF
2731 posts since 31 Aug, 2011

Post Fri Aug 31, 2018 3:11 am

Thanks, glad you enjoyed it. :)

Like i said, there is serious potential in SoftVoice, its just that the tech is very old by now and that it wasnt really made with singing in mind to begin with. (Its a TTS after all.) But if they would continue to develop it (make syllables sound more natural, etc.) im sure it could become a very usable voice generator even today especially if it was made available as a VST plugin with ALL parameters exposed for editing since this would allow you to design your voice from the ground up to give it exactly the character that you have in mind instead of being dependent on a handful of presets where all the important parameters that ultimately define the character of the voice are fixed and unalterable.

User avatar
wagtunes
KVRAF
19715 posts since 8 Oct, 2014

Post Fri Aug 31, 2018 3:29 am

ENV1 wrote:Thanks, glad you enjoyed it. :)

Like i said, there is serious potential in SoftVoice, its just that the tech is very old by now and that it wasnt really made with singing in mind to begin with. (Its a TTS after all.) But if they would continue to develop it (make syllables sound more natural, etc.) im sure it could become a very usable voice generator even today especially if it was made available as a VST plugin with ALL parameters exposed for editing since this would allow you to design your voice from the ground up to give it exactly the character that you have in mind instead of being dependent on a handful of presets where all the important parameters that ultimately define the character of the voice are fixed and unalterable.
Which is pretty much what I'm waiting for.

I have a feeling I've got a long wait.

ENV1
KVRAF
2731 posts since 31 Aug, 2011

Post Sat Sep 01, 2018 4:31 am

Youre probably right.

Even a non-VST version where at least all the voice parameters are exposed for editing would be great, but for that you would probably have to obtain a license and sorta 'roll your own'.

(Im pretty sure that Kraftwerk had a version that allowed them to go beyond presets and pitch and speed because some of the vox bits they used definitely sound like SoftVoice but not like any preset/pitch/speed combination that is possible with TalkIt or Willow Pond or ShitTalker or any of the other SoftVoice based apps that only expose Presets/Pitch/Speed. They also had the german syllable set too, which TalkIt has a slot for on the UI (the empty area below Spanish) and which is referenced in the DLL (TIGERM32.DLL), but not actually included in TalkIt for reasons unknown. So IMO it could well be that they had a custom version tailored to their needs, after all they even had hardware made specially for them so why not software too.)

User avatar
vurt
addled muppet weed
68538 posts since 26 Jan, 2003 from through the looking glass

Post Sat Sep 01, 2018 4:39 am

at that time too, what better advert for your skills?

oh you want my cv? listen to manmachine, i coded that voice.

you're hired!

ENV1
KVRAF
2731 posts since 31 Aug, 2011

Post Sun Sep 02, 2018 5:33 am

I think it would be cool even today. If there were a version thats freely accessible to everyone, like Talk It! which originally came with an addon-package for Windows 95 named 'Microsoft Plus! for Kids', with all parameters exposed for editing, im sure the tinkerers out there would come up with all sorts of cool and usable voices because technically SoftVoice is capable of way more than Hyper Male and Old Woman and Martian Robots.

But of course the ultimate coolness would be if they made a VST version where you can not only access all those voice parameters in order to design your own custom voices, but where you can also automate stuff via a word/phrase editor like Vocaloid has in order to modify important parameters directly during playback. This would allow you to really make it sing because youd be able to program things like pitch progressions dynamically and exactly as required. (And the software already has that capability, only its via textual input only.) Unfortunately its probably not going to happen because im sure it would be difficult to compete against apps like Vocaloid even if those are sample based and therefore rather limited with regard to voice characters. But then again, you never know...

DSmolken
KVRAF
1721 posts since 20 Sep, 2013 from Poland

Post Sun Sep 09, 2018 11:32 pm

fisherKing wrote:i know, for myself, i would never use any plugin, no matter how good, to replace a real singer.
I would, and I have, on a track with a very good human singer who wrote the song herself. And it wasn't to sound robotic. I just wanted to try a violin part as an aah vocal with legato instead, it sounded good, and why drag the singer from another city to a recording session for 30 seconds of aah when having a voice with a different timbre also makes things easier to separate in the mix? Neither she nor I saw a reason to do that, though for the next song she did record her own backing vocals, aahs, oos and all.

At some point as the needed vocal grows in length and complexity it becomes faster and more convenient to get a human to do it, but for small bits...

User avatar
Bansaw
KVRian
1202 posts since 19 Aug, 2008

Post Sat Oct 13, 2018 6:47 am

I got hold of this recently, enjoying trying to make it sound the way I want.
I can't get it to say "leave" - the full sentence is "to make you leave" but its coming out "to make you leeee"
(pic attached)
I even tried to record "vuh" elsewhere and I'l paste it on the end of leee but I can't get it to just give me a "v" without it giving me a "veeee" sound.

Any ideas?
You do not have the required permissions to view the files attached to this post.

User avatar
wagtunes
KVRAF
19715 posts since 8 Oct, 2014

Post Sat Oct 13, 2018 6:52 am

Bansaw wrote:
Sat Oct 13, 2018 6:47 am
I got hold of this recently, enjoying trying to make it sound the way I want.
I can't get it to say "leave" - the full sentence is "to make you leave" but its coming out "to make you leeee"
(pic attached)
I even tried to record "vuh" elsewhere and I'l paste it on the end of leee but I can't get it to just give me a "v" without it giving me a "veeee" sound.

Any ideas?
Well, that's either a bug or a very poorly implemented engine.

Suggestions?

See if eve works
See if leaf works

Try to narrow down where the bug is.

User avatar
Bansaw
KVRian
1202 posts since 19 Aug, 2008

Post Sat Oct 13, 2018 12:08 pm

wagtunes wrote:
Sat Oct 13, 2018 6:52 am
Well, that's either a bug or a very poorly implemented engine.
Suggestions?
See if eve works
See if leaf works
Try to narrow down where the bug is.
Thanks @wagtunes. Good tips, leaf was a touch better but I still got the "leeee" sound. I tried putting "voh" and it does give me the right v sound at the end of leave, but there is a slight gap between leve and voh, but I can probably edit this together in my DAW and pull in the gap.
You do not have the required permissions to view the files attached to this post.

User avatar
wagtunes
KVRAF
19715 posts since 8 Oct, 2014

Post Sat Oct 13, 2018 12:25 pm

My point in using "eve" was to see if it actually said "eve" and not "eeee". Is the software capable of putting the "V" sound at the end of a word?

If you say "love" does it say "love" or does it say "luhhh"

I'm trying to narrow down where the bug is? Is it "V" or is it the "EV" combination of letters?

Either way, it's pretty sad if this thing can't say leave and makes it totally unusable for me.

But like they say, you get what you pay for.

User avatar
Bansaw
KVRian
1202 posts since 19 Aug, 2008

Post Sat Oct 13, 2018 12:32 pm

wagtunes wrote:
Sat Oct 13, 2018 12:25 pm
If you say "love" does it say "love" or does it say "luhhh"
But like they say, you get what you pay for.
It does say "luhhh" actually. This is a bug. I might contact the developer and let him know.
Is Vocaloid a $100 better than SynthV though? You were pretty down on it on other thread I saw. ... ?

User avatar
wagtunes
KVRAF
19715 posts since 8 Oct, 2014

Post Sat Oct 13, 2018 12:48 pm

Bansaw wrote:
Sat Oct 13, 2018 12:32 pm
wagtunes wrote:
Sat Oct 13, 2018 12:25 pm
If you say "love" does it say "love" or does it say "luhhh"
But like they say, you get what you pay for.
It does say "luhhh" actually. This is a bug. I might contact the developer and let him know.
Is Vocaloid a $100 better than SynthV though? You were pretty down on it on other thread I saw. ... ?
My beef with V5 is the UI. It is a PITA to work with compared to the Cubase editor for V4 and that's working IN Cubase.

But quality wise as far as sound, there is no comparison. You don't have stupid bugs like the one above. It works flawlessly.

FWIW, you may not mind the UI if you've never worked with the V4 editor for Cubase. From what I understand, it's pretty standard for this kind of software. I'm just spoiled and that's why I hate using it. Has nothing to do with the sound, which, is actually a slight improvement over V4 with all the built in FX and things you can do with the vocals that you can't do in V4.

Pelicanomicon
KVRist
50 posts since 29 Sep, 2015

Post Sat Oct 13, 2018 11:44 pm

Bansaw wrote:
Sat Oct 13, 2018 6:47 am
I got hold of this recently, enjoying trying to make it sound the way I want.
I can't get it to say "leave" - the full sentence is "to make you leave" but its coming out "to make you leeee"
(pic attached)
I even tried to record "vuh" elsewhere and I'l paste it on the end of leee but I can't get it to just give me a "v" without it giving me a "veeee" sound.

Any ideas?
Downloaded the latest Windows version and copied the notes/lyrics according to your screenshot. Rendered at 80 & 180BPM and on my machine, SynthV clearly pronounces the "v" in "leave."

What happens if you use phonemes instead of phonetic approximations? (Highlight the problem note, go to Edit in the menu bar, then Convert to Phonemes.)

In Arpabet, "leave" should be ".l iy v"
However, Eleanor seems to pronounce this as ".l ih v" (rhymes with "give").
".l iy iy y v" sounds better.

Splitting the word into two notes also works: "[.l iy iy y][.v]"

Return to “Instruments”