Jump to content

Real Digitized Voices


Taz1004

Recommended Posts

I wanted to change the digital voice of the game.  The typical "Mission Success, all enemy forces have been eliminated" female digitized voice.  It sounds like Microsoft Zira which to me is the most unnatural one.  So I was looking to change it to Mark who I think is much more natural.  I looked it up in Mission Editor and it looks like these are not Text to Speech engine but recorded sound files?

 

Digital voice or "Text to Speech" has gotten a lot better and natural in recent years.  Siri, Google, Alexa, are all complete digital voices.

I wish to have actual "Text to Speech" engine implemented in DCS.  Not only more natural ones are available, we can finally have different voices for different missions.  There are different accents too.  UK, Australia, Canada...


Edited by Taz1004
  • Like 2
Link to comment
Share on other sites

On 1/11/2021 at 3:58 AM, Taz1004 said:

These are some examples of fully digitized voices today.  From Amazon Polly

 

Newscaster Voice

 

Conversational Voice

 

And most of them already have the API.

 

Wow that is impressive. Id like to see that on DCS...

Specs: Win10 64bits Pro, Intel i9-9900K | 32Go | RTX 2080 Ti | M.2 SSD 850go x2

Hardware: HTC Vive Pro + X56

Maps : Normandy + Assets | Gulf | Nevada

DCS Modules: FC3 | UH-1H | Mi-8MTV2 | A-10C | F/A-18C | Ka-50 | SuperCarrier | F-14A/B | F-5E | F-86F Sabre | MiG-15bis | Mig-19 | MiG-21bis | AV-8B | Fw 190 D-9 | SA342 | P-51D | Bf 109 K-4 | Spitfire LF Mk. IX | M-2000C | F-16C

 

Link to comment
Share on other sites

On 1/12/2021 at 12:55 AM, Cthulhus said:

 

Wow that is impressive. Id like to see that on DCS...

I disagree, they both sounded unnatural to me. A little bit clipped.

Granted, the tech is light years better than Max Headroom, but still

not quite there IMHO.

🇺🇦  SLAVA UKRAINI  🇺🇦

MoBo - ASUS 990FX R2 Sabertooth,     CPU - AMD FX 9590 @4.7Gb. No OC
RAM - GSkill RipJaws DDR3 32 Gb @2133 MHZ,   GPU - EVGA GeForce GTX 1660Ti 6Gb DDR5 OC'd, Core 180MHz, Memory 800MHz
Game drive - Samsung 980 M.2 EVO 1Tb SSD,    OS Drive - 860 EVO 500Gb SATA SSD, Win10 Pro 22H2

Controls - Thrustmaster T-Flight HOTAS X,   Monitor - LG 32" 1920 X 1080,   PSU - Prestige ATX-PR800W PSU

Link to comment
Share on other sites

15 minutes ago, rayrayblues said:

I disagree, they both sounded unnatural to me. A little bit clipped.

Granted, the tech is light years better than Max Headroom, but still

not quite there IMHO.

 

Really.  You think current in-game female digitized voice "Mission accomplished, you may RTB or engage targets of opportunity" is better?  I think Max Headroom was actually better than her.  Not surprising since Max Headroom actually was voice actor.

 

What's worse is those digitized voices in game are sound files.  People ask why training missions are always outdated.  It's because they have to re-record sound files whenever something changes.

 

Text to speech engine can also vary tones and expressions.  Sad, happy...  Doesn't have to be same voice all the time.


Edited by Taz1004
  • Like 1
Link to comment
Share on other sites

That's not what I said. I was referring to the two sound samples you provided. They are a vast improvement over what we have, but still sound a bit off to my ears.

I agree that we definitely need better voices and those samples would be just fine. All I meant was that they still sound like digital voices.

I was trying to be a little humorous with the Max Headroom comment,

I know he was voiced by Matt Frewer who did a great job of exaggerating a digital character. (digital, get it?)

I am a old blues and jazz musician. I have sensitive ears and I still believe that vinyl records sound better than CD's. 


Edited by rayrayblues
  • Like 1

🇺🇦  SLAVA UKRAINI  🇺🇦

MoBo - ASUS 990FX R2 Sabertooth,     CPU - AMD FX 9590 @4.7Gb. No OC
RAM - GSkill RipJaws DDR3 32 Gb @2133 MHZ,   GPU - EVGA GeForce GTX 1660Ti 6Gb DDR5 OC'd, Core 180MHz, Memory 800MHz
Game drive - Samsung 980 M.2 EVO 1Tb SSD,    OS Drive - 860 EVO 500Gb SATA SSD, Win10 Pro 22H2

Controls - Thrustmaster T-Flight HOTAS X,   Monitor - LG 32" 1920 X 1080,   PSU - Prestige ATX-PR800W PSU

Link to comment
Share on other sites

Text-to-Speech is sorely needed. Currently mission designers use pre-recorded sound files. That works only for missions that follow a rigid script. 

It doesn't work in dynamic missions. How for example would a mission call out the coordinates of a dynamically generated unit, whose spawning location is determined by live lua script and depends on the exact mission state. What is needed is a trigger.action.outTextToSpeech() in analogy to trigger.action.outText().

  • Like 3

[sIGPIC][/sIGPIC]

 

Intel Core I7 4820K @4.3 GHz, Asus P9X79 motherboard, 16 GB RAM @ 933 MHz, NVidia GTX 1070 with 8 GB VRAM, Windows 10 Pro

Link to comment
Share on other sites

I’ve never heard this digitized voice in the game. What is it?

i9-13900K @ 6.2GHz oc | ASUS ROG MAXIMUS Z790 HERO | 64GB DDR5 5600MHz | iCUE H150i Liquid CPU Cooler | 24GB GeForce RTX 4090 | Windows 11 Home | 2TB Samsung 980 PRO NVMe | Corsair RM1000x | LG 48GQ900-B 4K OLED Monitor | CH Fighterstick | Ch Pro Throttle | CH Pro Pedals | TrackIR 5

Link to comment
Share on other sites

10 hours ago, rayrayblues said:

I disagree, they both sounded unnatural to me. A little bit clipped.

Granted, the tech is light years better than Max Headroom, but still

not quite there IMHO.

Still better that what we have now? I don't like how numbers/indications are pronounced on DCS right now like that all is "disconnected". Not fluent.
 
English isn't my natural language so maybe I don't notice all the imperfection. But to me, it's better than what we have right now.
 
For now, the best I have seen so far, but again, perfection is what we have in the ATC of the last commercial SIM released in 2020. (Don't know if I can say the name?)

Edited by Cthulhus
  • Like 2

Specs: Win10 64bits Pro, Intel i9-9900K | 32Go | RTX 2080 Ti | M.2 SSD 850go x2

Hardware: HTC Vive Pro + X56

Maps : Normandy + Assets | Gulf | Nevada

DCS Modules: FC3 | UH-1H | Mi-8MTV2 | A-10C | F/A-18C | Ka-50 | SuperCarrier | F-14A/B | F-5E | F-86F Sabre | MiG-15bis | Mig-19 | MiG-21bis | AV-8B | Fw 190 D-9 | SA342 | P-51D | Bf 109 K-4 | Spitfire LF Mk. IX | M-2000C | F-16C

 

Link to comment
Share on other sites

  • 2 weeks later...

I agree.

 

I've just downloaded a few SP missions for the Gazelle, and even though I appreciate the written dialog in the missions, the immersion would have been increased immensly by having the dialog spoken out. I've tinkered with VoiceAttack, which have the functionality to read out text using Microsoft speech engine, even choosing between the different personas for different commands, very cool! Unfortunately Microsoft Speech Engine has a hard time understanding my Scandinavian English, so I seldom get any response - same as talking to my teenager now that I come to think of it... 

 

A command in the Mission Editor labeled: "Speak text", with an input field for text, a drop down menu for choosing voice, and perhaps dropdown for language. And maybe a global setting where you could choose which speech engine to use, and default language and voice. That would be sweet.

It might not be perfect, but it sure would be good enough in most cases. For the perfect delivery in very scripted missions, you would still have the existing option of using pre-recorded audio.

  • Like 1
Link to comment
Share on other sites

Not just better and more voice, but its mechanic is very much needed.  Right now, they're holding off on training missions on EA modules like F-16 because things are likely to change.  And whenever something change, they have to redo the voice overs.  Like in A10C-II there are some incorrect voices in training missions.  And recording voices alone is a lot of work but when you're revising just a portion, it's even more difficult because you have to match that voice to the previous recording.  Or have to re-record everything.  And then they have to sync it with the text.

 

With Text to Speech, you can revise anytime with none of these issues.

 

This is another Text-to-Speech example.  Again, not perfect.  But far better than what's currently in the game.  And just as OpenVR API gets updated, Text-to-Speech engine will get updated and improve over time.

 

 


Edited by Taz1004
Link to comment
Share on other sites

On 1/9/2021 at 1:53 AM, Taz1004 said:

Digital voice or "Text to Speech" has gotten a lot better and natural in recent years.  Siri, Google, Alexa, are all complete digital voices.

I wish to have actual "Text to Speech" engine implemented in DCS.  Not only more natural ones are available, we can finally have different voices for different missions.  There are different accents too.  UK, Australia, Canada...

 

 

Mostly agree. 5 years ago there were few nice open source voice engines that became very natural ones. The expensive commercial ones were little better, but not so much.

But the topic is very complex and challenging, why I think ED should go straight to Open Source versions, create from it the own module or program that they can offer for DCS users (can even be then with paid) with the download link for the modified sources.

 

As while it would be nice to have it IN the DCS, I think what we mainly need is just a way to generate the audio files for mission editor. So quickly write the text you want to get, copy-paste to external program, select some factors (gender, accent, language) and generate the wav files. And then get them automatically saved to proper place. From there then get them in DCS editor where the radio effects (distance, weather etc) are generated to that audio file.

 

As I do not see ED to come anywhere close in acceptable timeframe about the proper TTS sounding.

i7-8700k, 32GB 2666Mhz DDR4, 2x 2080S SLI 8GB, Oculus Rift S.

i7-8700k, 16GB 2666Mhz DDR4, 1080Ti 11GB, 27" 4K, 65" HDR 4K.

Link to comment
Share on other sites

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...