RT-Voice - Run-time text-to-speech solution

Hello Stefan. I have question.

It's a really good job to buy your assets.
I am making an android car game.
I have problem. Speaker.Speak --> Expensive. There is Lek.
I found what you wrote earlier. that
----------------------------------------------------------------------------------------------------------------------------------------------------------
Hello again

RTV hasn't a built-in cache; we tried this for a long time, but the requirements of our customers make this an extremely difficult task...

However, there is nothing holding you back from rolling your own solution. If you take a look at the API, you will see that the "Speak"-call has a parameter called "outputFile":
https://www.crosstales.com/media/da...ass_crosstales_1_1_r_t_voice_1_1_speaker.html

You can use those files to populate your cache. E.g. every text has an id, so you look up the file with its id; if it's there, play it. If not, call "Speak" with id as "outputFile".


I hope this helps you further.
----------------------------------------------------------------------------------------------------------------------------------------------------------
I want to do that. I tried all day, but I couldn't.
I don't know how to used That
Can you Help me Some Example source code Please

 

Hi

We're currently working on a port of our powerful caching system from DJ.
But this will need some time - please send me your invoice and I can give you access to the latest version as soon as it's ready.


So long,
Stefan

 

I think it is amazing you guys have kept up with this product for 5+ years. Well done

I'm curious if you have any recommendations on a library that does Speech to Text? We are thinking we would use your library for TTS but need a second library to do the reverse. Unless this asset has some support for that as well?

 

Hi ickydime

Thank you for the kind words!
If you're happy with the product, don't forget to write a review

I assume you mean "speech recognition"? RTV doesn't offer this functionality, but a partner publisher has a solution that may fit your needs:
https://assetstore.unity.com/packages/add-ons/machinelearning/google-cloud-speech-recognition-vr-ar-mobile-desktop-pro-72625

I hope this helps you further.


Cheers
Stefan

 

We bought your package and it works beautifully. One problem we have however is that if someone with a Samsung phone tries to change the voice/language of the TTs then his options will be different from another phone which has the google engine as his default?

Is there any way on android in RTVoice to set the engine to Google's and get its voices to use?

 

Hi

Unfortunately, Samsung uses (on some phones) their own TTS - the Google engine isn't even installed...
There are two oprions:

1. "force" your affected users to install the Google engine
2. Use an online service like AWS etc.

I hope this helps you further.


Cheers
Stefan

 

Do you mean that if it is installed then voices will appear regardless of the default engine?

 

@Laubenberger I checked this and if I changed my preferred engine in screen reader settings to Google, then I have many options, otherwise on Samsung I have English and another language.

As I looked, the android text to speech engine allows you to send your preferred engine in the constructor. Do you allow us to do that in RTVoice in any way? If not is it possible to add something to the next version so we can try to use Google's engine if available ?

 

Hello again

RTV uses the default engine on purpose; however we test if we can expose the engines and let the user choose the desired one. But I can't promise anything right now.


Cheers
Stefan

 

@Stefan-Laubenberger Many thanks
That would be great in general and help us save costs by not having to use the cloud providers most of the times

 

I'm trying to get Mandarin characters to work. I can get Pinyin to work. I'm using AWS so it may be related to that or perhaps the way the characters are being sent to AWS (most likely).

Should Mandarin characters work or just Pinyin?

 

Hello there!
Tried to change the voice from the default to female voice by using

Code (CSharp):

1. Speaker.VoiceForGender(Crosstales.RTVoice.Model.Enum.Gender.FEMALE)

Its always playing with the default voice.

Tried also the Test-Drive on the RTVoice prefab... it is working fine and showing the femail voice as well.

 

Hello again

We added support for all installed TTS engines under Android. If you like to have access to the latest beta, please send us an email with the invoice.


Cheers
Stefan

 

Hi ickydime

Unfortunately, I don't know much about the various Chinese writing styles and signs.
Is it possible that you can provide some example texts? This way I would be able to try it on my own.


Cheers
Stefan

 

Hi Olkan

Are you waiting for the "OnVoicesReady"-callback before using the "Speak"-method?
You can find an example in chapter 5.2.7 of the documentation:
https://www.crosstales.com/media/data/assets/rtvoice/RTVoice-doc.pdf


Cheers
Stefan

 

Hi,
Quick questions...I'm using windows 10 and will create offline app. So is there any license fee to use windows TTS ? Can I use this plugin offline?
Thank you

 

Hi

No, there is no license fee for all built-in voices under Windows.
It may defer for 3rd party voices from Cereproc etc.

I hope this answers your question.


Cheers
Stefan

 

I've been having an issue in iOS. I can't seem to stop the AVAudioSession without it throwing this warning:

[AVAudioSession setActive:withOptions:error:]: Deactivating an audio session that has running I/O. All I/O should be stopped or paused prior to deactivating the audio session.

What I have tried:
Speaker.Silence()
Speaker.Pause()
SpeechText.Silence()
AudioSource.Stop()
AudioListener.pause()

This occurs with Native iOS voices, Amazon Polly and Google.

Has anyone run into this and solved it?

 

Hi Jason

Unfortunately, I never saw this error before.
However, since you experience the error with the non-native providers, like AWS and Google, I bluntly blame Unity for this issue
The reason is that other providers are using only Unity-based audio stuff, so I currently can't imagine, what else would cause the problem...
Which Unity version are you using?


Cheers
Stefan

 

Hi Sefan,

I'm on 2018.4.22 LTS.

Well I guess I can live with that warning for now.

On a related note I was also having issues getting Amazon and Google voices to start up again after De-Activating the AVAudioSession. But I finally found a work-around if anyone else is having this issue...

The issue was that after re-enabling the AVAudioSession (for example to start Audio Ducking of other apps) the cloud providers would not start speaking again. The solution I found is to call UnityEngine.AudioSettings.Reset(AudioConfiguration);

I could not determine what actually was causing this issue and I'm pretty sure this solution works by accident. It also has the side effect of interrupting any other music playing so it's not perfect but at least the voices start up again.

 

Hi,
I just installed RT Voice PRO on my Mac, and I receive 2 errors:

Assets/Plugins/crosstales/RTVoice/Scripts/Provider/VoiceProviderMacOS.cs(31,54): error CS1729: 'BaseVoiceProvider' does not contain a constructor that takes 1 arguments

Assets/Plugins/crosstales/RTVoice/Scripts/Provider/VoiceProviderMacOS.cs(41,13): error CS0103: The name 'speakerObj' does not exist in the current context

What I have to do to solve these issues??

Massimo

 

Hi Massimo

Thank you for reporting this issue!

We fixed the problem in the current beta version; please send us the invoice via email and we will give you access to the fixed version.


So long,
Stefan

 

Hi @Stefan-Laubenberger, I bought RT-Voice, and really like it.
I am experimenting to build a Mac OSX app with basic RT-Voice in Windows env. My RT-Voice's app in Windows is working the way I wanted, and would like to build it for Mac OS X. When I changed my Build Setting's Target Platform to Mac OS X, I get these 3 errors:
Assets\Plugins\crosstales\RTVoice\Scripts\Provider\VoiceProviderMacOS.cs(31,54): error CS1729: 'BaseVoiceProvider' does not contain a constructor that takes 1 arguments
Assets\Plugins\crosstales\RTVoice\Scripts\Provider\VoiceProviderMacOS.cs(41,13): error CS0103: The name 'speakerObj' does not exist in the current context
Assets\Plugins\crosstales\RTVoice\Scripts\Speaker.cs(2230,33): error CS7036: There is no argument given that corresponds to the required formal parameter 'obj' of 'VoiceProviderMacOS.VoiceProviderMacOS(MonoBehaviour)'

I searched in the forum thread and Google, but no luck finding answer.

I am using Unity 2020.1

Is there any tutorial or instruction for creating Cross Platform app?

Thanks!

 

Hi SirCodeWarrior

Please update to the latest version from the store.


Cheers
Stefan

 

Thank you very much, @Stefan-Laubenberger for your fast response! It works!

 

first impressions are WOW! this is amazing! we will see if my brain can keep up with implementing it.

i posted on your tutorial youtube video about what I am trying to do

I am trying to build a twitch chat tts, i have bought this and twitch irc by voodoo.

Everything is working very well.

I will do a video in a few days of what i am trying to do. This is on linux cinnamon mint 20.

I am trying to integrate and implement cereproc natural voices.

I have their sdk and a sample voice called heather.voice. I have no idea how to integrate cereproc voices into this workflow. I will try to figure it out and bug them if I can interface with marytts.

 

Hi Chris

Just let me know if you need any assistance.


Cheers
Stefan

 

Problem 1: how to i integrate crosstales rt voice pro and twitch irc so that
it can read what comes up in the chat box and also what i type and send. see video below to see the problem visually.
This should be very easy to do.

problem 2: how do i integrate cereproc with unity / rt voice pro / mary tts.
i see that rt voice pro has an alternative option to mary tts and espeeak. How do i plug cereproc into that?

problem 3: when i compile the project and run it outside the editor i do not get any voices coming out.
is this something to do with the error?

error MissingReferenceException: The object of type 'AudioClip' has been destroyed but you are still trying to access it.
Your script should either check if it is null or you should not destroy the object.

here is a wip video.

 

Hello again

Here are the answers to your questions:

1. It's along video and we don't know Twitch IRC... Can you point me to the exact problem? You can simply call Speaker.Speak("text") whenever you like.
2. You are using Cereproc under Linux, right? If that's the case, I'm afraid you would need your own custom provider. I can help you guide in the right direction. Personally, I would recommend using a cloud service like AWS Polly, Azure or Google on Linux - I know it's hard but they sound best
   
3. What's your provider for the build? MaryTTS or eSpeak?

Cheers
Stefan

 

1. i will try to use Speaker.Speak("text") to solve this problem thanks for the tip!
2. I will try to bug Cereproc to help me integrate with you.
3. For the build i was using mary tts because the voices are better using your test server. i will test with espeak and also with a local mary tts server.

 

some problems:

1. I added the line Speaker.speak(ChatText.text);
to twitch irc and get an error.

//Receive message from server
void OnServerMessage(string message)
{
ChatText.text += "<b>SERVER:</b> " + message + "\n";
Speaker.speak(ChatText.text);
Debug.Log(message);
}

ERROR
Assets/TwitchIRC/Example/Scripts/TwitchIrcExample.cs(43,9): error CS0103: The name 'Speaker' does not exist in the current context

normally i would declare something along the lines of using rtvoicepro; // to access functionality

normally i would create a public game object and drag that in to enable crossover functionality.

3. espeak works, connecting to your test server does not work on a standalone build, how do i access my local mary tts server?

i can access my local server marytts in firefox via http://localhost:59125/
When iput that in as the address i get this error.

System.Net.WebException: Could not find a part of the path "/home/useri5/july 2020 tts 1/localhost:59125:59125/voices". ---> System.Net.WebException: Could not find a part of the path "/home/useri5/july 2020 tts 1/localhost:59125:59125/voices". ---> System.IO.DirectoryNotFoundException: Could not find a part of the path "/home/useri5/july 2020 tts 1/localhost:59125:59125/voices".
at System.IO.FileStream..ctor (System.String path, System.IO.FileMode mode, System.IO.FileAccess access, System.IO.FileShare share, System.Int32 bufferSize, System.Boolean anonymous, System.IO.FileOptions options) [0x00164] in <fb001e01371b4adca20013e0ac763896>:0
at System.IO.FileStream..ctor (System.String path, System.IO.FileMode mode, System.IO.FileAccess access, System.IO.FileShare share, System.Int32 bufferSize, System.IO.FileOptions options) [0x00000] in <fb001e01371b4adca20013e0ac763896>:0
at System.IO.FileStream..ctor (System.String path, System.IO.FileMode mode, System.IO.FileAccess access, System.IO.FileShare share, System.Int32 bufferSize, System.Boolean useAsync) [0x00000] in <fb001e01371b4adca20013e0ac763896>:0
at System.Net.FileWebStream..ctor (System.Net.FileWebRequest request, System.String path, System.IO.FileMode mode, System.IO.FileAccess access, System.IO.FileShare sharing, System.Int32 length, System.Boolean async) [0x00000] in <14e3453b740b4bd690e8d4e5a013a715>:0
at (wrapper remoting-invoke-with-check) System.Net.FileWebStream..ctor(System.Net.FileWebRequest,string,System.IO.FileMode,System.IO.FileAccess,System.IO.FileShare,int,bool)
at System.Net.FileWebResponse..ctor (System.Net.FileWebRequest request, System.Uri uri, System.IO.FileAccess access, System.Boolean asyncHint) [0x00026] in <14e3453b740b4bd690e8d4e5a013a715>:0
--- End of inner exception stack trace ---
at System.Net.FileWebResponse..ctor (System.Net.FileWebRequest request, System.Uri uri, System.IO.FileAccess access, System.Boolean asyncHint) [0x000a5] in <14e3453b740b4bd690e8d4e5a013a715>:0
at (wrapper remoting-invoke-with-check) System.Net.FileWebResponse..ctor(System.Net.FileWebRequest,System.Uri,System.IO.FileAccess,bool)
at System.Net.FileWebRequest.GetResponseCallback (System.Object state) [0x00074] in <14e3453b740b4bd690e8d4e5a013a715>:0
--- End of inner exception stack trace ---
at System.Net.WebClient.OpenRead (System.Uri address) [0x000a1] in <14e3453b740b4bd690e8d4e5a013a715>:0
at System.Net.WebClient.OpenRead (System.String address) [0x00016] in <14e3453b740b4bd690e8d4e5a013a715>:0
at (wrapper remoting-invoke-with-check) System.Net.WebClient.OpenRead(string)
at Crosstales.RTVoice.Provider.VoiceProviderMary.getVoicesInEditor () [0x000b1] in /home/useri5/july 2020 tts 1/Assets/Plugins/crosstales/RTVoice/Scripts/Provider/VoiceProviderMary.cs:611
UnityEngine.Debug:LogError(Object)
Crosstales.RTVoice.Provider.VoiceProviderMary:getVoicesInEditor() (at Assets/Plugins/crosstales/RTVoice/Scripts/Provider/VoiceProviderMary.cs:665)
Crosstales.RTVoice.Provider.VoiceProviderMary:.ctor(String, Int32, String, String) (at Assets/Plugins/crosstales/RTVoice/Scripts/Provider/VoiceProviderMary.cs:45)
Crosstales.RTVoice.Speaker:initProvider() (at Assets/Plugins/crosstales/RTVoice/Scripts/Speaker.cs:2205)
Crosstales.RTVoice.Speaker:ReloadProvider() (at Assets/Plugins/crosstales/RTVoice/Scripts/Speaker.cs:2076)
Crosstales.RTVoice.EditorExtension.SpeakerEditor:OnInspectorGUI() (at Assets/Plugins/crosstales/RTVoice/Editor/Extension/SpeakerEditor.cs:182)
UnityEngine.GUIUtilityrocessEvent(Int32, IntPtr) (at /home/builduser/buildslave/unity/build/Modules/IMGUI/GUIUtility.cs:197)

How do i access marytts as a local server?
do i have to put it somewhere specific for the apllication to see it?

 

I solved problem number 1 by putting twitchirc into the namespace Crosstales.RTVoice.Demo

namespace Crosstales.RTVoice.Demo

{

   public class TwitchIrcExample : MonoBehaviour

  {

     public Text VoiceText;

    Speaker.Speak(VoiceText.text);

  }

}

 

Hello again

Sorry for the delay, but I was traveling and couldn't answer sooner.

Here are some details to your questions:

1) As mentioned at the beginning of chapter 5 in the docs, you have to use "using Crosstales.RTVoice;" in your scripts to gain access to our classes

3) You have to separate the url and the port:


I hope this helps you further.


Cheers
Stefan

 

Thanks for your patience and support it really is helping! i have been looking into cereproc (because i love their accented voices) and they offer a cloud based service much like azure / aws / google could you help me with integrating that into rtvoice pro? https://api.cerevoice.com/v2/ it has a low use free tier for developers who own the sdk. (which costs 60 pounds)

 

Hi Chris

The CereProc-API looks fairly easy and similar to the MaryTTS-provider.
We have planned to add it as an official provider in the future, but I can't give you an ETA atm.

If you're in a hurry and want to do it yourself, take a look at "VoiceProviderExample.cs" - you could create a copy and fill in the modified code from the MaryTTS-provider.


So long,
Stefan

 

Ok, my current problem is that all the speech kicks in at once, how do i pause it speaking after each line
i put in do and while from the dialog example but it didnt know what playing or running was in the current context.

Code (CSharp):

1. do
2.                   {
3.                      yield return null;
4.                   } while (!playingB && Running);
5.  
6.                   //wait until played
7.                   do
8.                   {
9.                      yield return null;
10.                   } while (playingB && Running);
11.  

Assets/TwitchIRC/Example/Scripts/TwitchIrcExample.cs(86,29): error CS0103: The name 'playing' does not exist in the current context

How do i make sure it pauses after each message rather saying it all at once? i will just put a few seconds pause in for now.

 

Hi

You have to wait for the "OnSpeakComplete"-callback before you fire-up the next line.
Please take a look at the example in chapter 5.2.7 of the documentation:
https://www.crosstales.com/media/data/assets/rtvoice/RTVoice-doc.pdf

I hope this helps you further.


Cheers
Stefan

 

Problem = five instant server messages at the same time speaking at the same time

Attempted solution 1: results in losing all but the first message.
Q. Do I have to put them all into a string table then read them off?

Code (CSharp):

1. using System;
2. using System.Collections.Generic;
3. using System.Collections;
4. using Irc;
5. using UnityEngine;
6. using UnityEngine.UI
7.  
8.  
9. private bool isSpeaking =false;
10.  
11.  
12.   void OnServerMessage(string message)
13. {
14.  
15.      if (!isSpeaking)
16.             {
17.               isSpeaking = true;
18.               StartCoroutine ("Pause1",message);
19.               Debug.Log("OnServerMessage " + message);
20.             }
21.  
22. }
23.  
24.  
25. IEnumerator Pause1(string message1)
26.         {
27.        // if (!isSpeaking)
28.        //    {
29.             isSpeaking =true;
30.             ChatText.text += "<b>SERVER:</b> " + message1 + "\n";
31.             VoiceText.text = "<b> " + message1 + ":</b> " + "\n";
32.             ListenText.text = VoiceText.text;
33.             Speaker.Speak(VoiceText.text);
34.  
35.             yield return new WaitForSeconds (3);          
36.             Debug.Log("wait is over");
37.             isSpeaking = false;
38.         //   }
39.         }
40.  

 

I would create a queue of strings and enqueue the new messages and use a separate loop to dequeue them.

 

Current problem I cant get subsequent messages to be spoken from the server.
UPDATE: I think i might have been overthinking this as by putting Speaker.Speak(message); in various places in the irc code seems to solve the problem without having to go too deep into rtvoice pro.


When i use direct chat to talk to stream it all works fine. Multiple chat inputs = multiple voice output.

Getting messages back from server it doesnt give me any voice. its a bit weird.

At the moment I am using Speaker.Speak(message);
this works for first messages but subsequent messages are not spoken.

Do I need to release the Speaker? for future speech?

Can you show me how to check if voice is ready. Then check if Speaker if finished before moving on.

I have had a look at the dialog example.

I am struggling to follow the code.

Would it be a while loop?
While (!Speaker.OnSpeakComplete)
{
//wait till Speaker is finished?
}

is it something to do with these? is it something to do with Speaker.OnSpeakComplete?

Speaker.OnVoicesReady
Speaker.OnSpeakStart
Speaker.OnSpeakComplete

I understand that I should be using onEnable() and OnDisable() to initialise and shutdown event listeners. i dont really understand what we are doing with speakStartMethod and speakCompleteMethod

Code (CSharp):

1. public void OnEnable()
2.       {
3.          // Subscribe event listeners
4.          Speaker.OnSpeakStart += speakStartMethod;
5.          Speaker.OnSpeakComplete += speakCompleteMethod;
6.       }
7.  
8. public void OnDisable()
9.       {
10.          // Unsubscribe event listeners
11.          Speaker.OnSpeakStart -= speakStartMethod;
12.          Speaker.OnSpeakComplete -= speakCompleteMethod;
13.       }
14.  
15. // I dont really understand why we using model or wrapper
16.  
17. private void speakStartMethod(Model.Wrapper wrapper)
18.   {
19.          if (wrapper.Uid.Equals(uidSpeakerA))
20.     {
21.  
22.     }
23. }
24.  
25. private void speakCompleteMethod(Model.Wrapper wrapper)
26.   {
27.          if (wrapper.Uid.Equals(uidSpeakerA))
28.     {
29.  
30.     }
31. }

 

Hello again

It would maybe better if we discuss this issue via email.

However, for your case, I think you only need "Speaker.OnVoicesReady".
You should start calling "Speaker.Speak" after OnVoicesReady is called.
I would not rely on the "Dialog"-example; this is what I would do:

1. Use the Queue-class and store all incoming messages.
2. Use "OnVoicesReady" to start a co-routine to process the messages
3. Inside the routine, use a while loop and dequeue message after message (but only if "Speaker.isSpeaking == false"). Don't forget to yield with something like "WaitForSeconds(0.5f)".

That's basically all you have to do.


Cheers
Stefan

 

Another problem I have is when i build an unity project in linux then run it i cannot seem to check if it is connecting to the internet. For example i make it with your test mary tts server but i dont think when i build it standalone that it connects to anything. Is there some setting i need to make sure it can actually connect?

 

You can't use our MaryTTS-server in builds; you have to use your own server or request a test-account on our server (by sending us your invoice).

 

Thanks for that information i thought it was a problem on my end.

I meant to suggest something, I use atavism and they provide a ubuntu virtualbox linux server. Maybe thats something you could do provide a virutal box marytts server for people to use that would distance reliance on your test server. So should i request a test account with you? I will try to setup a test server on another machine.

 

Hello and thank you for your asset.
I have followed the basic setup on this video:

, it works well when playing in Unity Editor. But I am working on a Oculus Quest app and it does not work when I press the button. I havent set a voice name to have the one by default on Oculus (I hope there's one).
I get this error :

Code (CSharp):

1. MissingReferenceException: The object of type 'AudioClip' has been destroyed but you are still trying to access it.
2. Your script should either check if it is null or you should not destroy the object.

I don't know how to solve this, is there someting to set in RTVoice ->Android Settings ->Engine or something else ?

 

Hello again

Please send us your invoice and desired user name to gain access to our MaryTTS-test server.


Cheers
Stefan

 

Hi

Can you please add "Debug.Log(Speaker.Engines.CTDump());"? This should display all available speech engines in Logcat.
Then choose an engine and set it with "Speaker.AndroidSpeechEngine".

I hope this helps you further.


Cheers
Stefan

 

Another question.
How do i change the parameters of the marytts voice to make it default to a faster and higher pitch?

using just the rtvoice prefab.

My problem is that i have created a real time chat tts so i dont want to use pre written text.

P.S.
I set up a local virtual box on a separate computer on my lan to test stuff out. I dont want marytts working on ths same computer as the main program (localhost)

I have setup a virtual box running mint 20 installed marytts used ifconfig to get ip address, applied bridged network to virtualbox i put the ip address in to rtvoice as the marytts server. It all seems to be working pretty good.

 

You can use the "Speaker.Speak"-call like this:

Code (CSharp):

1. string yourText = "Test"; //your texts
2. string yourDesiredVoice = "cmu-rms-hsmm"; //name of the MaryTTS voice
3. float yourRate = 1.2f; //can be between 0 - 3
4. float yourPitch= 1.5f; //can be between 0 - 2
5.  
6. Speaker.Speak(yourText, null, Speaker.VoiceForName(yourDesiredVoice), true, yourRate, yourPitch);

I hope this helps you further.


Cheers
Stefan

 

Thank you Stefan,

I have added "

using Crosstales;

" to be able to use

CTDump()

Result is com.svox.pico
But the problem was the following: it was trying to load a fr-FR voice, I had this error in the Android Logact window :

Error SVOX Pico Engine Failed to load textana resource for fr-FR

So I have set the language in English and it works:

public string Culture = "en";

Speaker.Speak(Text.text, null, Speaker.VoiceForCulture(Culture));