Suggestion: Make Radio Com sound a bit more Radio like
-
Actually not really, you can script these type of audio operations using sox command line tool or dll. My point was more along the lines of fidelity.
Then again it’s why I made this application to do this via text to speech. Wth the right voices installed (good ones are payware) the entire process is automated.
https://www.benchmarksims.org/forum/showthread.php?33733-TlkTool-GUI
-
Better method would be to have cleaner, updated voice clips .
Sure, but that’s 30000 samples to either re-record, with 14 different voice actors (12 pilots 2 ATCs) –> hugely impractical, or have an automated voice do --> so the authenticity you might gain in radio effects, you will lose out on voice intonation.
-
You can listen to the results from some payware (and non payware) voices this example video…
The text strings are stored in the F4Talk95v1-0-0.csv file so linking them to the fragfile.xml is trivial.
Since we know both the ID and strings it’s completely automated.
Of course TTS has it’s limitations, but you can also edit the strings to have the proper pronunciation, and this also only has to be edited once for all voices.
-
As far as I know, TTS is seriously considered … But still need to wait a bit more for technology to be more “democratised” (and enhanced for emotions/intonations). For the moment, free softwares are still too luch robotic for real satisfying results.
-
So IVC uses canned samples for some things – the blocked sound for instance. Aside from that though, the other effects, including things like 400Hz AC hum, are generated via signal processing routines on the fly. In part that is to allow things like atmospheric and altitude effects to be applied dynamically so that you get the effect of more severe signal impairment at longer ranges and when flying low in clutter. Trying to do the same thing via a static edit of the individual sound samples will of necessity not deliver as satisfying result since you won’t get the dynamic attenuation that IVC’s implementation delivers.
AI sounds are processed a little with some parameterized distortion by the main game code but the effect is weak and kind of arbitrary by comparison. I’ve considered porting the DSP effects from IVC to the main code to use those on the AI speech as well but formats aren’t compatible and so forth making it not exactly straightforward either. On top of that there is some serious discussion about changing the way that AI speech is curated in future – perhaps not on-demand TTS but rather using TTS to generate the equivalent of frags and what have you. Given that, it didn’t seem like a good time at the moment to put a great deal of energy into this area.
-
Getting rid of the classic radio voices would be a major nostalgia killer for me
-
Is it possible to get a sample?
Sure! As for now I have a generic background noise with open and close click , I think IVC effects are more close to what we hear in some RL videos but they are good anyway.
My 2 cents, I think only flight radio calls needs some effects. Maybe even AWACS and tanker… ATC is good as it is now.
-
Getting rid of the classic radio voices would be a major nostalgia killer for me
+1
Hopefully it will never happen.
-
+1
Hopefully it will never happen.
Yeah, still missing the ‘Guns,guns,guns…’ from the original.
-
Hopefully it will never happen.
It will happens.
I mean, if we like to go forward and consider more about AWACS comms’ JTAC etc … there is not ither solution since current available frags do not cover our needs any more. This is also why CeckIn freq is not yet really implemented in 3.34 … And why ATC and AI calls are not 100% accurate in terms. Tweaking exisitng frags has a its limits. -
It will happens.
I mean, if we like to go forward and consider more about AWACS comms’ JTAC etc … there is not ither solution since current available frags do not cover our needs any more. This is also why CeckIn freq is not yet really implemented in 3.34 … And why ATC and AI calls are not 100% accurate in terms. Tweaking exisitng frags has a its limits.OK, I will understand and appreciate expansion in area of communication (more speeches), but I really want to have current voices with Falcon 4.0 origin kept in future releases.
-
People want to keep 8 bit 22khz radio voices? LMAO…
Are you guys also running 640x480 in 3D for that “classic” experience?
-
OK, I will understand and appreciate expansion in area of communication (more speeches), but I really want to have current voices with Falcon 4.0 origin kept in future releases.
Imagine if we mix TTS an real voice frag! … for a same pilot … and differences on each words!? … The day we will switch into full TTS, it will just be impossible to keep old voices. Unless you can fine original actors and ask them to records new frags/sentences with the same tone they did 20y ago …
For instance : you guys want the L16 … how to make AIs saying “Hook” “Zap” “Markpoint” “Track Number” … etc … if the frags doesn’t exist!
Do you understand why we can’t (existing frags vs non existing frags)? -
There are some good TTS nowadays ( mostly payware ), the problem is, if you want to do a “good” job, you have to work on a single sentence , doing a mass porting will make things worst.
I use this for DCS voiceovers. They work very well but they are not perfect.
-
There are some good TTS nowadays ( mostly payware ), the problem is, if you want to do a “good” job, you have to work on a single sentence , doing a mass porting will make things worst.
I use this for DCS voiceovers. They work very well but they are not perfect.
From quick review most strings were said with the correct pronunciation with payware TTS.
All voices share the same text strings so it only has to be edited once per string in question which is do-able as most are ok.
Many are just single words, letters and numbers, not everything is a phrase.
The proper solution for BMS would be to integrate a TTS engine into the code, and just read text strings externally. Recording TTS to a WAV clip is kinda a hack to get them in BMS.
-
Recording TTS to a WAV clip is kinda a hack to get them in BMS.
A superficial look at the problem would suggest this is correct.
However, practically all of the AI speech is parameterized such that what you hear is assembled from pieces parts rather than recorded as a sentence in it’s entirety. There is a good reason why the eval/frag system that Microprose chose has lasted so long…replacing it with something better is quite tricky.
TTS engines are only recently starting to get to the level of quality that would make literal on-the-fly assembly of strings for pronunciation in real time viable (for things other than ATIS that use stilted TTS in real life also; in certain places anyway). And as things stand today most of the alternatives that look most promising are commercial which puts them effectively out of reach for BMS to use.
-
A superficial look at the problem would suggest this is correct.
However, practically all of the AI speech is parameterized such that what you hear is assembled from pieces parts rather than recorded as a sentence in it’s entirety. There is a good reason why the eval/frag system that Microprose chose has lasted so long…replacing it with something better is quite tricky.
TTS engines are only recently starting to get to the level of quality that would make literal on-the-fly assembly of strings for pronunciation in real time viable (for things other than ATIS that use stilted TTS in real life also; in certain places anyway). And as things stand today most of the alternatives that look most promising are commercial which puts them effectively out of reach for BMS to use.
Have you read my replies in this thread?
I’m well aware of how the voice is strung together, as I needed that info to make the application for this task.
What I was trying to say before you made a quick assumption is that having TTS as wav files makes no sense pragmatically.
Wasn’t saying the current system is bad, it makes sense for recorded wav voices.
-
Have you read my replies in this thread?
Yes.
What I was trying to say before you made a quick assumption is that having TTS as wav files makes no sense pragmatically.
I know what you are trying to say and I disagree. That happens sometimes when reasonable people debate
-
Yes.
I know what you are trying to say and I disagree. That happens sometimes when reasonable people debate
Well we agree to disagree, at least I do
Might also be an option to do real voice recordings but not use so many variants, ie. 2 male 2 female (or people as Justin Trudeau would say)
As mentioned, BMS will probably never implement a payware TTS engine (understandably) so using WAV clips is the obvious answer.
To reply directly to the OPs initial question, I’ve already attempted to apply compression and some “radio” efx (bandpass, static, noise) globally to all wav clips and the end result didn’t sound any better in game. If the engine was set to use at least 16bit / 44khz samples then these effects would be much more audible if applied directly to the source.
-
Sure! As for now I have a generic background noise with open and close click , I think IVC effects are more close to what we hear in some RL videos but they are good anyway.
My 2 cents, I think only flight radio calls needs some effects. Maybe even AWACS and tanker… ATC is good as it is now.
This are 2 examples, I might review my thoughts but I think it can be done using IVC effects.
https://www.dropbox.com/s/necirko2f0q24yh/examples.rar?dl=0- IVC effects are in wav files so I’ve used them ( can be found in IVC folder )
HQ-noise-48kHz.wav
mic_in.wav
mic_out.wav
opencircuit.wav-
I’ve added this effects to original files with AUDACITY ( can be done massively to all files )
-
BMS just link wav files to each other to create phrases so , to obtain a similar method to IVC , it can be done like that
ch 1 - (open mic wav) + wav 1 + wav 2 + wav 3 + wav 4 + ( close mic wav )
ch 2 - –----------------------Open circuit wav ------------------------------------
ch 3 - ------------------------HQ-noise-48kHz.wav--------------------------------the only problem is to repeat background waves and cut them when original phrase is ended - can it be done?