Where is xVASynth.exe file?

[liuronghui]

Thank you for such an exciting project.

"To start, double click the xVASynth.exe file, and make sure to click Allow"

Where is xVASynth.exe file?

[DanRuta]

It's in the packaged/compiled release, over here: https://github.com/DanRuta/xVA-Synth/releases

I'll make sure to add a note, specifying this. Thanks for pointing that out.

[liuronghui]

Maybe I should download the Release version first and run it. I can't compile and run the desired results. At the beginning, my local machine Python version was 3.7, executing PIP install -r requirements.txt command error.Then, it changed to the 3.6 version to solve this problem, but it still failed to run the effect.No audio files can be exported.

[DanRuta]

That's strange.

Is there an audio file in the output folder (or in the output/femalecommander folder)? How are you running this? From npm start? If so, try opening the console (Ctrl + Shift + I), and please let me know if there's any error.

By the looks of your screenshot, the error is likely not in the python code, but if you check out the server.log file, there may be a message there, if there was an error. Do let me know if you find something.

The release version just removes the need for node or python (as it is compiled).

Side note: You'll probably need some more text, to generate something successfully.

[liuronghui]

Thank you for your reply

[DanRuta]

Hmm.. it would seem there might be an issue with the python code. With a clear log file, loading Delphine and generating that phrase should log out the following:

2018-10-15 12:21:40,156 - New session
2018-10-15 12:25:14,721 - POST /loadModel
2018-10-15 12:25:14,721 - {'outputs': 10, 'model': './models/skyrim/delphine', 'cmudict': False}
2018-10-15 12:26:52,292 - POST /synthesize
2018-10-15 12:26:52,292 - {'sequence': '50,36,47,35,64,47,50,42,64,35,48,41,31,45,32,31,64,46,32,49,32,41,47,52,64,29,36,39,39,36,42,41,64,52,48,28,41,64,56,64,29,36,39,39,36,42,41,57,64,36,41,64,50,32,28,39,47,35,58,64,37,28,30,38,64,40,28,55,46,64,33,28,40,36,39,52,64,29,32,30,28,40,32,64,47,35,32,64,45,36,30,35,32,46,47,64,36,41,64,30,35,36,41,28,1', 'outfile': './/output/temp-2045491735710041.wav'}

But it seems that either the server is not starting up, or the requests are not getting to it.

Have you tried running the compiled, release version? If so, are you running into similar issues?

Another thing you might try is running python server.py, and checking out http://localhost:8008/ in a browser, to see if you can see the [DEBUG] Get request for / message. If you can't, there must be an issue with the server start-up, and if so, there may be errors logged to the console.

Another thing I can see in your screenshot, unless this is an artifact, is that there is a gap in the file path for the audio file. You have file:///C:/xVA-Synt output/temp....., whereas I'd expect there to be something like file:///C:/xVA-Synth/output/temp...... I suspect this might be the source of the issue, if the model loading and synthesis actually worked ok (which seems to be the case, given the UI state). If it is, I'm unsure what could have caused this. Perhaps you could console.log the path variable, and the tempFileLocation variable on line 281 in script.js?

I'm unable to replicate the issue, but perhaps replacing line 281 with the following might help?

const tempFileLocation = `${path}/output/temp-${Math.random().toString().split(".")[1]}.wav`

[liuronghui]

http://morefunreadingproduct.oss-cn-shanghai.aliyuncs.com/219825606261608450/server/audio/2018-010-16/DetWY2ywMM-It's very nice of you to invite me.wav http://morefunreadingproduct.oss-cn-shanghai.aliyuncs.com/219825606261608450/server/audio/2018-010-16/DetWY2ywMM-it would seem there might be an issue with the python code. With a clear log file, loading Delphine and generating that phrase should log out the following.wav

http://morefunreadingproduct.oss-cn-shanghai.aliyuncs.com/219825606261608450/server/audio/2018-010-16/DetWY2ywMM-commented 6 hours ago.wav

Thankfully, it's already running, I've made a serious mistake by putting the code on an 8008-port web server instead of running it as server.py

But the generated audio is very strange. How should I improve it? I want to make it into Chinese speech synthesis,I have an idea to make a AI telemarketing system, Through technology like you, statements into audio, and make an interface service called by external system.

[DanRuta]

Ah, good to hear it runs! Those are some pretty unfortunate samples, though.

If you'd like to improve the quality, you can check out some of the tips on the readme.

However, judging by what idea you're describing, you'd be much better off using some models trained on actual speech datasets, not something that was hacked together with audio files from video games 😀.

I'm unsure if you've checked out the repos for the underlying technologies, but I'd advise you to go check them out instead, if you haven't already:

tacotron: (used in this project) https://github.com/keithito/tacotron tacotron-2 (I didn't have much luck with this one, yet, but you might): https://github.com/Rayhane-mamah/Tacotron-2

If you'd like to synthesise Chinese speech, you'd need to train a model using a Chinese data set. Perhaps you can follow the conversation here, for something to use: https://github.com/keithito/tacotron/issues/118

[FightingBlaze]

It's in the packaged/compiled release, over here: https://github.com/DanRuta/xVA-Synth/releases

I'll make sure to add a note, specifying this. Thanks for pointing that out.

Ok, I must be blind, because its nowhere in the 1.0 version. Unless its a separate file I need to download somewhere else?

[DanRuta]

I removed the builds from GitHub. Please get the tool from one of the Nexus pages. Although, at this moment in time, those are a little behind the code in this repo. I am due to push out a large update soon.

One of the pages is here (there is/will be one for most games supported): https://www.nexusmods.com/skyrimspecialedition/mods/44184?tab=files

[FightingBlaze]

Ok thanks man :)

[Felina-Lain]

I removed the builds from GitHub. Please get the tool from one of the Nexus pages. Although, at this moment in time, those are a little behind the code in this repo. I am due to push out a large update soon.

One of the pages is here (there is/will be one for most games supported): https://www.nexusmods.com/skyrimspecialedition/mods/44184?tab=files

Which game is the one that you update the files for in priority? I found the mod in both Witcher3 and Skyrim, but patch versions varied. Obviously I'm downloading the latest (v2.0.5 patch), but I'm wondering if there is yet another version I missed in another game on Nexus

[DanRuta]

The Skyrim page is the most popular, and the one I update first, normally. The upload process for nexus is very slow and sometimes difficult to work with, so I sometimes take longer to update all of them. I'm working on some things to change how the update process will work, but for now, Skyrim's the main one.

[Felina-Lain]

The Skyrim page is the most popular, and the one I update first, normally. The upload process for nexus is very slow and sometimes difficult to work with, so I sometimes take longer to update all of them. I'm working on some things to change how the update process will work, but for now, Skyrim's the main one.

Thanks, I'll track this one then! :)