What's new

Making believable voices: plugins & process?

creativeforge

the plumber
Moderator
So let's say you don't have the budget to hire a voice actor for a personal project. Modifying your own voice could potentially work, I assume?

What plugins and process would yield good, great or awesome results?

The project I'm thinking of is a reading of poetry, so digital voices out there really don't work. And my voice could use an edge to make it more dramatic...

[I hear this often on FM radio, they have the voice pitched down and apply effects on it. That. Then there is the "anonymous interviewee" we see on TV, to preserve their anonymity they garbled the voice with pitch and it sounds horrible - not that.]

Thanks!
 
I would hire an actor or a singer. I have worked with a guy who has the most amazing baritone voice I've ever heard. Speaks French and English. He's American but I've never asked him to do an English / Irish / non-American accent.

Let me know; a lot of singers are looking for work, I expect.
 
What kind of fees are they looking for?

I don't know. He's got an amazingly powerful, rich, and interesting voice. He sings all over the world with the Los Angeles Master Chorale, so he wouldn't be free. If it's for television or a film, that will certainly cost something more meaningful, but if it's not a commercial project, or it's spec, or something modest, I expect you can work something out.
 
I'll echo the option for hiring someone, if possible.

That said, zplane's elastique is quite good. You can test it out using the fully-functioning demo version of Reaper using the ReaPitch plugin. You can change the pitch and formant values to change the sound of your voice.

You can also try out Antare's Throat EVO, which has a demo but requires an iLok account.

By way of explanation, pitched sounds are being generated by passing glottal pulses produced by your vocal cords through your vocal tract. By adjusting the position of your tongue, you divide the tract into a series of chambers. Each chambers create resonances - called formants - which is how we create distinctive phonetic sounds. The position of the formants is basically a function of the length of your vocal tract.

For example, the position of the formants for an average male /EE/ sound are located at 270, 2300, and 3000Hz. Since these formants are a function of the length of the resonating chambers in the vocal tract, their position doesn't change when the pitch changes.

Pitch shifting by only speeding up or slowing down the sound shifts not only the frequency of the voice, but the frequency of the formants as well. That effectively changes the perceived length of the vocal tract, which is where the "chipmunk" or "giant" sounds come from.

To prevent this, pitch shifting software automatically readjusts the position of the formants so they don't change as the frequency changes. This allows you to change the pitch without altering the perceived size of the speaker's vocal tract.

ReaTune (and most other pitch-correcting software) allows you to do the opposite as well: move the position of the formants without altering the pitch, changing the perceived size of the vocal tract.

There are other factors that can be adjusted as well, such breathiness, tenseness, and so on. Some programs basically reconstruct the vocal signal from scratch, and allow you to replace the glottal pulse with a synthetic one. Throat EVO is one of the programs that allows you to modify these parameters.

Of course, you can modify the quality of your glottal pulse yourself, without any software. ;)

Depending on your needs, if you "create a character" with a different cadence than your own speech, alter some of your speaking patterns, and shift the formants and pitch to taste, you might be able to create an effective different voice.

But if you expect the software to be able to do most of that, I think you'll be disappointed by the results.
 
Surely there are Youtube tutorials on how to EQ/manipulate voice over? Maybe one of those will help?

I still think @David Cuny is right -- it's probably not going to be as cool as a live performer.
 
Hi there
Of course a native voice-over artist is the best but gives a lot of recording hassle...reflections first, kind of microphone, sharp esses etc.
iZotope’s Nectar 3 is a very cool and affordable vst plug-in which can solve
a lot of recording problems and even better it with compression, pitchcorrection
and other essential fx’s.
It also has nice voice-over preset folder !
 
For grins, I had a go at a using synthetic voice "Will" voice from Natural Readers at a speed of -2. I then modified some phrases in Reaper, and drew in some new cadences using Waves Tune.

Some bits, like "hollow men", were particularly resistant to editing.

Still, an interesting experiment.
 

Attachments

  • The Hollow Men (Portion).mp3
    5.2 MB · Views: 37
Interesting, if you speed the result up, does it make it better? Natural Reader was one of the sites I went to. But all their intonations are more for presentations, not audiobooks or poetry.
 
Interesting, if you speed the result up, does it make it better?
Yes, I picked that tempo because I liked the "dramatic" phrasing it gave at that speed. I could render it out at a faster tempo, but that adds another pitch shifter, and it creates some funky artifacts.

Obviously, the closer the initial version is to what you want, the less editing you'd need to do to get to the final result. So if I wanted it faster, I should have chosen a faster reading speed.

If I'd spent more time experimenting, I could have solved some problems that wouldn't have needed editing. For example, this wasn't read the way I wanted:

This is the cactus land.

The phonetic spelling works much better, not just for the pronunciation, but the phrasing as well:

This is the kaktiss land.

Similarly, each of these phrases gets a different performance:

Here, there is nothing.
Here there is nothing.
And here, there is nothing.
And here - there is nothing.
Here? There is nothing.


The dash and comma effect durations differently. The question mark is useful for extending the duration of the word - it's easier to change the pitch trajectory than to stretch a word.

I used Waves Tune to modify the pitches, because it's easy to draw in a new pitch trajectory with the pencil tool. Unfortunately, it's also got a maddeningly small UI, which makes it frustrating to work with.

You really need a pitch correction tool that allows you to draw smooth curves, because the spoken pitch trajectory isn't at all like a sung trajectory.

Which means you need to buy and learn another new tool. Joy! ;)

I still think a human reader could do a much better job, but it is possible to get not terrible results synthetic voice. :)
 
Last edited:
So let's say you don't have the budget to hire a voice actor for a personal project. Modifying your own voice could potentially work, I assume?

What plugins and process would yield good, great or awesome results?

The only one known to me that gives moderately decent results is Melodyne.
With a spoken voice you could try to moderately change the pitch and/or the formants.
However, the range where it does work is rather limited.
(typically in the range of a minor third).

Of course the usual suspects (EQ and compressor) will also help to get a better voice.
I'd try with an EQ finding the body of your voice (somewhere around 120-200 Hz) and moderately boost it. Gives more weight to what you say! :)
 
+1 for Melodyne. I'm still climbing the ladder to the ultimate package.

Any of these automated voices sounds like I've stumbled upon an Anonymous video about questionable subject matter.

I did find a South African female voice that was nearly spot-on. I had an hour of fun making my friend say horrifying things.

The bottom line is that the human voice is something we hear all the time. Formants and all that help, but our brains can tell when something just isn't right. I like this conversation. I believe automated solutions are getting awfully close. Give me a Scottish accent to play with; I'll lose a day of productivity laughing myself into convulsions.

Greg
 
Thank you all for your suggestions and comments! I see the most organic way to go is actually to hire a narrator... :)

Or use my own voice (French accent) but it's not as dramatic as a British (or Irish) accent.
 
You should try DAVID by IRCAM forumnet. AFAIK, it's available for free.

Also, check out the following free options:
ANGUS https://forum.ircam.fr/projects/detail/angus/
C.L.E.E.S.E. https://forum.ircam.fr/projects/detail/cleese/

SuperVP Trax is probably the closest to what you are looking for. You need a subscription to forumnet (about 200 USD), though.

Flux Audio offers the same algorithms found in SuperVP Trax with a more fancy interface and much higher price tag (same applies to SPAT, btw - which you can even get for free from IRCAM directly). Of course, Flux's solution offers plugins for DAWs.
https://forum.ircam.fr/projects/detail/supervp-trax/

Good luck!
 
Hi all, so here is an update!

Thank you again for the very helpful advice you shared, which I followed. I was fortunate enough to connect with a musician in the UK who connected me three other really awesome musicians who all know each other and work together.

I ended up hiring 3 different drummers, and auditioned 5 narrators before finding the right ones for the project. I also hired a bass player and a drummer, and a female vocalist. I took care of the piano and keys, VI and effects, and the whole editing and production.

Now I think I'm about done the mixing part but am looking for professional ears to give their impressions and comments. I posted a request for "beta listening" here. If you could help?


Thank you in advance!

André
 
Top Bottom