Another method to downgrade TTS voices to radio quality with low CPU load

[roland_lfor 245]

Hi,

 

I first tried the solution found here, but was not happy with it for several reasons:

1. The solution does not work all the time you start it. It looks like there is like a stochastic behavior (and I don't like that  )
2. The CPU load on my system reach 6%, and more strange the CPU load stays @ 6% aven after the end of TTS sentences from P2A.
3. I was unable to tune my radio sound as I wanted to with Voxal Voice Changer

So I spend some time over Internet to find an alternative solution, and discovered the world of VST Audio plugins.

 

Many freeware VST plugins can be found over Internet, but you first need a VST host.

I've chosen Minihost Modular because it's powerful, light, propose x86 or x64, and I like the graphical editing, wiring from a module to another with mouse.

 

Then I looked for a VST filter which could change the TTS voices to radio like and finally kept Izotope Vinyl module which doing a great job.

But to modify ONLY the TTS voices coming from P2A, you need a virtual cable - see below the download link.

 

So here is quick list to build your own:

1. Install Virtual Cable
2. Install Izotope Vinyl VST filter
3. Install Minihost Modular
4. Start P2A, in Speech settings, change the Speech output device to CABLE Input (VB-Audio virtual cable)
5. Start Minihost Modular
6. In Preferences, set Audio Device Type to DirectSound, set Input to CABLE Output (VB-Audio virtual cable)
7. In Preferences, Plugin tab, you have to Scan & verify to list available VST plugin before to use it (check where the Vinyl plugin is installed and if needed add the new folder before scanning. Close Preference dialog
8. In the main interface, right click->Add plugin->iZotope Vinyl
9. With mouse, add a wire between left yellow AudioIn pin to left input pin of Vinyl box, and add another wire between Vinyl output pin to Audio Out pin of Minihost Modular
10. Double click on Vinyl box to adjust you settings, you can play TTS from P2A to test them on the fly.
11. Once you are satisfied, save your graph, you are done.

Here is a view when it's ready:

 

 

On Vinyl plugin I only play with the following settings:

 

And the good thing is my CPU load is around 2%, but now only during TTS sentences.

 

Download links:

 

 

>>> Link to download the Virtual Cable from VB Audio Software

>>> Link to download the Minihost Modular

>>> Link to download the izotope Vynil filter

 

 

Notice that those components are freeware, but for the 2 last you need to register before download.

If some of you want to try and have some problem, I can give some help.

 

Hope you will like it  :smile:

[RobdeVries 28]

Hi Roland,

Thanks for your tutorial. It's fun and it works.
Now each time I want to have radio like voices I need to start minihost modular correct?
Too bad it does not have the static interference..

 

 

These are my settings. I added a little bit of mechanical and electrical noise.

 

Thanks very much!
Rob

[roland_lfor 245]

Happy to see it's useful for someone :smile:

 

The only problem is that it's no possible to launch automatically Minihost Modular with the configuration file as a command line argument...

[RobdeVries 28]

Well it's beta so you we could ask for an extra feature

[Tesch Macher 1]

hi, just stumbled up on this as i was looking to modify the TTS voices to sound more real

except that instead MiniHost from image-line.com , i went with SaviHost http://www.hermannseib.com/english/savihost.htm   , exact same principle and much more easier to use.

 

[roland_lfor 245]

Hi, Assuming we can use VST filters, is it possible to run Savihost from a command line with the setup name as an option?

This is what I miss with MiniHost.

[mkurver 6]

Thanks for all your very helpfull input on how to downgrade the TTS voices so they sound more realistic in the sim.

It made me think about an even simpler solution. Basically all we need is some kind of AM-radio filter on the channel where your headset audio-device is plugged in.

I found that the free Equalizer APO works very well. It is modular so you can easily add effects and also some VSP plugins.

 

When installing Equalizer APO the installer asks you for which audio device you want the equalizer to work. I choose my USB Headset.

In Equalizer APO I only modified the 15-band equalizer. Putting all channels below 300hz and above 2000hz on zero.

Everything in between I played with what sounds best (400hz on 15, 630hz on 18, 1000hz on 9 and 1600hz on 10)

Besides the Equalizer module I added the Filter module with a Low-pass filter on 1750Hz.

 

(please note: I am a complete noob when it comes to sound editing...so I just played around until I found the TTS voices sounded credible)

 

I started up Pilot2Atc and tested some voices in the configurator and compared them to real world ATC.

When I was happy with the settings I closed Equalizer APO and next time you start your pc the filter is automatically applied. No extra actions required, all sounds via that sound device are now modified.

I havent noticed any performance drop when using this software.

 

PS: the downside is that this software always modifies the sound coming through the audio device you have set up.

I believe there is a way to bypass this by assigning a Equalizer-profile to a specific application via one of the many Equalizer APO mods available like this one. 

I havent tried these yest because I am using an aviation headset dedicated to flightsim.

 

Let me know if this is helpfull for you!

good luck!

 

 

[Bt103504 30]

Greetings Gents,

Would any of you be willing to post a quick video or sound clip of your results? I might be interested in using one of your processes outlined above to enhance the immersion of my TTS voices. @mkurver 

[peloto 213]

On 10/30/2017 at 10:20 PM, Tesch Macher said:

hi, just stumbled up on this as i was looking to modify the TTS voices to sound more real

except that instead MiniHost from image-line.com , i went with SaviHost http://www.hermannseib.com/english/savihost.htm   , exact same principle and much more easier to use.

 

Could you or someone make a mini tutorial for noobs in this?

Why is this not implemented natively by Pilot2ATC? It would be a great feature.

Edited August 24, 2019 by peloto

[peloto 213]

On 8/24/2019 at 4:28 PM, peloto said:

Could you or someone make a mini tutorial for noobs in this?

Ah, ok, is the same like the first post but different program. Sorry. I am going to try it step by step as described up, to make it work.

[peloto 213]

I was able to do it all the same with the first post, is always the background sound emitting and no only when the ATC speaks? If so, is a bit annoying to have it enabled.

[Dave-Pilot2ATC 870]

Here's a link to a video I found which explains how to do this using Donationware/Free software.

https://www.bing.com/videos/search?q=pilot2atc+sound&docid=608007037916022521&mid=76D1ECF1EE06589E2F3676D1ECF1EE06589E2F36&view=detail&FORM=VIRE

Dave

[Theonlyjez 0]

I used this method and had great results..

I downloaded all the English speaking win 10 voices and put them through the mixer and Equalizer and assigned them all to a different ATC position.

All sounding very realistic now..

 

Thanks Dave for a great product..

[motishow 123]

an't find Minihost Modular link, anyone an share it pls? thanks

[Roman Design 344]

I think there is a much better method that should work:

1. In recent Windows 10 you can specify audio device PER APP. Like this.

2. You can use free Voicemeeter Banana mixer to create/use a separate virtual audio device for P2ATC app. It has built-in EQ. Just cut lows and highs to the max and boost mids and it should emulate radio spectrum. It won't affect main sound or FSFS sound then, coming out on default audio device. 

3. I believe Pilot2ATC has a setting for "Radio background sound" for a suitable static file. Together EQ and static should be enough for very convincing radio sound. No need for any VST plugins.

Needs to be tested as latency can become an issue, if it slows down the main audio output, but latency should be lower than using VST plugins anyway.

Also, if using chatter files, EQ should not affect them much, as those frequencies are already absent. If it does - you cut too much!

Edited April 29, 2021 by Roman Design