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

If you use non standard (=system) TTS voices, you will have to deliver them somehow...
There are imho three possibilities:

1. Let the user download the needed voices (not very user-friendly and error prone)
   
2. Include and install the needed voices with your application. But be very careful: you have to read the license agreements for those voices. Afaik it can be very costly to obtain a single voice from a vendor for this purpose, since every user of your app has all the voices "forever" and can use them outside of your scope. There are some free/oss voices available but their quality is often low. And again, you have to check the license for distribution etc.
   
3. Create all "fixed" speeches with your desired voices as audio files and stick with a standard voice from Windows and Mac for "interactive" (runtime) speak-parts. You can tweak them with pitch, rate and audio filters to get some variance.

I personally would choose number 3.

I hope this helps.


Cheers
Stefan

 

Thanks for this great product but I am confused.

How doI use this for Anroid?
I use Playmaker and RTvoice pro, newest RTvoice, win10 and newest Unity5

thanks

 

Hi eagleeyez

Hmm, I'm a bit confused too

You can use RTV exactly as for all other platforms.
Did you see the "3rd party"-folder inside our asset? It contains a PlayMaker-package with some demo scenes.
You can set the desired voice in the "Speak"-action, here an example from the "PlayMaker_Demo"-scene:



I hope this helps you further.

Have a nice weekend!


Cheers
Stefan

 

Hi and thanks for your reply.

In the screenshot you sent me, you can see

RtVoice name windows = Microsoft David

RTvoice name IOS = Alex

But by

RTVoice name android = nothing here.

But I built the example mobile scene to my s7, at first it did not work as it complained about manifest files. So I updated the Java SDK to version 23 and it worked.

I was surprised that it worked, where do I find the name for the voice in Android and how to set this in RTVoice? Do I just have to write for example. Alex, when in Android an Alex TTS is installed?

Thanks again

 

You're welcome!

You can build a release with the "Mobile"-scene inside the "Demo"-folder - there you get all the available voices.
And yes, just type the name of the voice (as in the "Mobile"-scene) and you're fine.

The reason for the empty Android-field are the various Android-versions out there - in nearly every major version the names change...

 

Thanks, I've understood it now, I was somehow standing on a hose pipe and now it is clear.

 

No problem, enjoy your weekend!

 

Thanks for the advice!

I have just run into another problem with the text to speech. I have successfully made the text to speech save to a file and update every few plays. However, I have a windows computer and some of my other group members use macs. For some reason the voices aren't loading for them, and they continuously get the following error:
Win32Exception: ApplicationName='say', CommandLine=' -r 157 -o "/var/folders/83/f3lm05md4fv4cyfbgfmwbxx80000gn/T/d3699beb-a270-4945-b39d-272e4f233fc9.aiff"

Is there a problem with what I'm doing or is there a way around this error?

The only line of code with the text to speech that I have is:
Speaker.Speak(announcement, weather, Speaker.VoiceForCulture("en", 0), false, .9f, 1, Application.dataPath + "/Sounds/Weather.wav", .7f);

 

Ok, I think it's a problem with the "say"-command. Which macOS-version are you using?
Can you please start the "Terminal"-app under macOS and enter the following command: which say
Then copy the result path to the clipboard and open the "ConfigLoader.cs" inside your project.
Add the following line at the end of the constructor:

Code (CSharp):

1. Constants.TTS_MACOS = "path to say command";
2. //Constants.TTS_MACOS = "/usr/bin/say"; //the result from our macOS

Then restart Unity and try it again.
If you encounter any warnings/errors in the console, please post the complete messages.

Another point - don't add an extension for the generate file, it should look like this:

Code (CSharp):

1. Speaker.Speak(announcement, weather, Speaker.VoiceForCulture("en", 0), false, .9f, 1, Application.dataPath + "/Sounds/Weather", .7f);

The reason: Windows writes "wav"-files and macOS "aiff"-files and the extension is handled by RTV.

Please let me know if this works.


So long,
Stefan

 

Thanks so much for your help again! I finally got one of my team members to test the fix, and it works now! I also changed the extension, so thank you for that advice.

 

You are welcome, I'm happy it's working for you!

Just a question - how exactly did your Mac-team fix the issue? Can you post the code and macOS-version?
This would allow me to improve RTV in a future version.

Thank you!


Cheers
Stefan

 

Hello,

I have RTVoice working great on PC/Mac in a fairly empty scene. Unfortunately on iOS and Android it speaks very slowly and is a bit hard to understand as it is playing so slow. Any thoughts on what I should check to get it to play normal speed on mobile? The scene seems to run at a good frame rate on mobile so it seems to be independent of performance.

Here is a video to explain: https://goo.gl/photos/NGPkpK1tWEkwzyQj7

 

Hi CollosalChris

Unfortunately, I can't run your video - it doesn't start.
Anyway, here are some suggestions on how to speed-up the talking under mobile:

1. Speak- and SpeakNative: change the rate value
2. Speak- and SpeakNative: change the pitch value
3. Speak: change the pitch of the audio source

Here are more details.

I hope this helps, please let me know.


Cheers
Stefan

 

Works great Stefan. Thanks!

 

You're welcome and have a nice day!


So long,
Stefan

 

Hello, I have an issue with RT-Voice Pro, after building (as stand alone app) my app to the Windows platform, sometimes (i can't understand reasons why it's happens) RT-Voice didn't work. After rebooting, rebuilding, etc. (some times few times) build become work correctly, it's reproduced at two computers where we develop apps. Maybe you can help determinate reason - why it's happens?

I attach a log file.

 

Hi Maxim

It looks like a weird problem, I never heard of it before...
The problem is that the static variable go (=gameobject) of RTV is somehow null (which should never be the case).

Let's test something: add RTV only in your first scene (GameScene?) and delete it in all other scenes. Then create a build and test it.
Please let me know if you still experience those issues.


Cheers
Stefan

 

Hello Stefan,
As I understand, RTV script automaticaly add RTV object to the game scene, so if i try to remove RTV gameobject to the scene and then open this scene againe - RTV go - will be create automaticaly... is it possible to disable "autocreation"?

 

Yes, that's not a problem. Just open the "Configuration" and disable "Prefab auto-load":



I hope this helps!


So long,
Stefan

 

Hello Stefan.
I have a question, is there a way to get a generated audio file to use it without playback?

 

Hi Oleksii

Yes, you can store the generated audio files as wav (Windows) or aiff (Mac), e.g:

Code (CSharp):

1. Speaker.Speak("<TEXT>", <AUDIOSOURCE>, <VOICE>, false, 1f, 1, "<PATH_TO_YOUR_FILE_WITHOUT_EXTENSION>");

See also the API:
http://www.crosstales.com/en/assets/rtvoice/api/class_crosstales_1_1_r_t_voice_1_1_speaker.html

Or you can use the "SpeechText"-component. Enable "Generate Audio File" and store your voices directly in the Editor (by clicking "Test-Drive"):



I hope this helps!

Have a nice weekend!


Cheers
Stefan

 

Hello Stefan, thanks for information, looks like your's script execute some 3rd party application, and our antivirus think so this is not safety and reject executing. After disabling antivirus everithing begin work correct, hope this issue will not reproduce anymore.

 

Hi Maxim

We use our own TTS-wrapper executable under Windows and sometimes a anti-virus engine displays a false positive.
We would never deliver any software that causes harm to our customers!

Here is the metadefender-result from the actual exe:

Cheers
Stefan

 

Hi - We're having a problem building an Android APK. I see the following in the console:

...Temp/StagingArea/android-libraries/RTVoiceAndroidBridge/AndroidManifest.xml:3] Main manifest has <uses-sdk android:targetSdkVersion='23'> but library uses targetSdkVersion='24'

Any ideas how to fix this?

Other than that it seems to work well for our purposes.
-Bob Gonsalves

 

Hi Bob

Did you update Android Studio to the latest version and which Unity-version are you using?


Cheers
Stefan

 

Thanks - Unity is at 5.4.2f2. I will try to update Studio, though it's not clear to me why that would have any impact. Or is it the Android SDK that needs updating? (We pretty much stopped direct Android development once we switched to Unity). Thankfully it does build on Unity Cloud Build.

 

Imho Android Studio includes the SDK, or not?
Anyway, please update it on your local machine and it should work.
I'm happy that the Cloud Build works at least.

 

Unfortunately updating the SDK (and Studio) did not solve this. I'm assuming there is a manifest inside your .aar file, is that correct. Where else could I look for manifests?

 

You can manually provide your own manifest file; it looks like right now you are simply using the manifest that Unity automatically generates if you don't provide one manually. When an Android app is built, Unity puts the generated manifest file in the Temp folder at StagingArea/AndroidManifest.xml - copy that file to edit it. Provide one manually by putting it in Plugins/Android/.

I hope this helps.

 

Thank you, that is a good suggestion and will be useful in the future. I ended up reinstalling the Android SDK and bumping up my Java version too.

 

I'm glad it's helpful for you!
If you like our asset/support, please rate us at the store. Or even better, write a review

So long,
Stefan

 

We've started playing the Unity 5.6 beta and see some errors thrown on the Mac/Editor. I haven't tried to track this down further, but:

Speaker.Speak(Locale.GetString("hello"),null,Voice(),true,1.0f,1.0f); and
Speaker.Speak(soundName,audioSource,<voice>,false,<rate>,1.0f,soundPath);

both throw
MissingMethodException: Method not found: "UnityEngine.WWW.GetAudioClip()"
in VoiceProviderMacOS+MoveNext()

which seems really strange.

 

We only support official released Unity-versions.

Anyway, I took a look at the API from 5.6 and there is no longer the "GetAudioClip"-method - it's now an extension method and we have to recompile it with 5.6.


So long,
Stefan


Edit:
This is fixed since the 2.6.0-release

 

Thank you for looking into this. We're already using SpeakNative in other parts of the code and will switch over to that. Hopefully this is just an oversight in the new API.

 

Stefan thanks for good tool!
Is it possible insert in text any special command (like [[slnc 1000]] for Silence ) for make beep or any system sound?

 

Hi

That's a great suggestion and I added it to our "Experimental"-list

Thank you and have a nice day!


Cheers
Stefan

 

Stefan, I updated to 2.5.2 this evening and got the following error and no speech to text voice can be heard on my MAC (it was working on 2.5.1):

No voice for culture 'en' found! Speaking with the default voice!
UnityEngine.Debug:LogWarning(Object)
Crosstales.RTVoice.Speaker:‭‪‌‮‮‫‍‍‫‫‬‬‮‏‫‮‍‮‏‬‫‎‌‬‫‍‏‮(Object)
Crosstales.RTVoice.Speaker:VoiceForCulture(String, Int32)
GameBoardlayNarrationSpeech(String) (at Assets/Scripts/GameBoard.cs:701)
GameBoard:SpeechToggle() (at Assets/Scripts/GameBoard.cs:655)
UnityEngine.EventSystems.EventSystem:Update()

Do I need to configure anything post the update?

 

Stefan,

I noticed that the RTVoice prefab wasn't loaded. Problem fixed! However, I am trying to select the serena voice on my MAC. I have selected Serena as the voice default for English (United Kiogdom), and the following SpeechNative call a strange english speaking voice is heard rather than serena:

Speaker.SpeakNative(speechText, Speaker.VoiceForCulture("en"), currentRate, currentNarrationVolume, currentPitch);

Serena is listed as the fourth voice in the MAC system preferences speech to text list for English (United Kingdom) but only serena is selected.

How can I guarantee that serena is always used? A similar chalenge for Windows and iOS platforms for my game.

 

Hi Ezzra

Use the following method:

Crosstales.RTVoice.Speaker.VoiceForName ( string name )


E.g.:

Code (CSharp):

1. Speaker.SpeakNative(speechText, Speaker.VoiceForName("Serena") , currentRate, currentNarrationVolume, currentPitch);

You have to find similar voices for all supported platforms and handle it yourself.
Here is an example for your own code:

Code (CSharp):

1. /// <summary>Name of the RT-Voice under Windows.</summary>
2. [Tooltip("Name of the RT-Voice under Windows.")]
3. public string RTVoiceNameWindows = "Microsoft David Desktop";
4.  
5. /// <summary>Name of the RT-Voice under macOS.</summary>
6. [Tooltip("Name of the RT-Voice under macOS.")]
7. public string RTVoiceNameMac = "Alex";
8.  
9. /// <summary>Name of the RT-Voice under Android.</summary>
10. [Tooltip("Name of the RT-Voice under Android.")]
11. public string RTVoiceNameAndroid = string.Empty;
12.  
13. /// <summary>Name of the RT-Voice under iOS.</summary>
14. [Tooltip("Name of the RT-Voice under iOS.")]
15. public string RTVoiceNameIOS = "Daniel";
16.  
17. /// <summary>Name of the RT-Voice.</summary>
18. public string RTVoiceName
19. {
20.     get
21.     {
22.         string result = null;
23.  
24.         if (Util.Helper.isWindowsPlatform)
25.         {
26.             result = RTVoiceNameWindows;
27.         }
28.         else if (Util.Helper.isMacOSPlatform)
29.         {
30.             result = RTVoiceNameMac;
31.         }
32.         else if (Util.Helper.isAndroidPlatform)
33.         {
34.             result = RTVoiceNameAndroid;
35.         }
36.         else
37.         {
38.             result = RTVoiceNameIOS;
39.         }
40.  
41.         return result;
42.     }
43. }
44.  
45.  
46. // Call example
47. Speaker.SpeakNative(speechText, Speaker.VoiceForName(RTVoiceName), currentRate, currentNarrationVolume, currentPitch);

I hope this helps, please let me know.


Cheers
Stefan

 

Hello Stefan, thanks for making such an awesome asset available. It's a shame that Unity hasn't prioritized accessibility but that's a discussion for another topic.

I purchased RT-Voice PRO 2.5.2 and I am able to hook into VoiceOver on my dev machine (I'm using OSX version 10.11). For my game I'd like the player to simply be able to mouse over Unity UI text elements and have them spoken. I've written a simple class (HoverSpeakText) that seems to do this and it might be a helpful start for anyone looking for the same kind of functionality.

Code (CSharp):

1. using System.Text.RegularExpressions;
2. using Crosstales.RTVoice;
3. using UnityEngine;
4. using UnityEngine.EventSystems;
5. using UnityEngine.UI;
6.  
7. public class HoverSpeakText : MonoBehaviour, IPointerEnterHandler
8. {
9.     const string HTML_TAG_PATTERN = "<.*?>";
10.     const string SPEECH_PREFERENCE_KEY = "kHoverSpeech";
11.     Text label;
12.     string fixedText;
13.  
14.     void Awake()
15.     {
16.         this.label = GetComponent<Text>();
17.     }
18.  
19.     void OnEnable()
20.     {
21.         if (!Speaker.isTTSAvailable || this.label == null)
22.         {
23.             gameObject.SetActive(false);
24.         }
25.     }
26.  
27.     #region IPointerEnterHandler implementation
28.  
29.     public void OnPointerEnter(PointerEventData eventData)
30.     {
31.         if (PlayerPrefs.GetInt(SPEECH_PREFERENCE_KEY, 0) == 1)
32.         {
33.             Speaker.Silence();
34.             if (this.fixedText == null)
35.             {
36.                 this.fixedText = Regex.Replace(this.label.text, HTML_TAG_PATTERN, string.Empty);
37.             }
38.             Speaker.SpeakNative(fixedText);
39.         }
40.     }
41.  
42.     #endregion
43. }

As you can see I'm using rich text on Text objects and I use regex to remove any html style tags from the text content before calling SpeakNative. Am I duplicating functionality that RT-Voice already handles? This seems like it might be a global setting, but I haven't found it in digging around.

Also I'm experiencing the console warning: "'Voice' or 'Voice.Name' is null! Using the OS 'default' voice."
My use case is to use the 'default voice' that the player has selected on their system to do speech, or as a fallback any default voice. I'd rather not specify cultures or have to expose these settings to the player. I can comment out the warning, but it seems as if I'm missing something in regards to setting up RT-Voice.

Finally, I build my game on all the desktop platforms Unity provides (Linux 64, Linux Universal, Linux, OSX Intel 64, OSX Intel, OSX Universal, Windows 64, Windows) and I would definitely benefit from having RT-Voice work on Linux. But my question is about 'Enforce 32bit voices', do I need to enable this setting for Windows 32 builds? Sorry if there is no correlation between the setting and the Unity Build setting, but it I really don't know what 'Enforce 32bit voices' does. Is it a setting that the player needs to enable if there are issues?

Thanks again for making this awesome asset!

 

Hi Feaver1968

Thank you for your post!

We didn't include a "clean text"-method since we don't know what the user wants to speak with our tool. Probably the tags are an intentional part of the speech...
Anyway, a little helper method or flag to solve this issue could be useful, so I think about a solution for a future release.

About the warning: afaik there isn't a default voice in the "say" command under macOS (and every other platform), so you should specify one for every platform and culture your app supports. Then handle it in combination with "Application.systemLanguage" and "Speaker.VoiceForName" (if you need an example just let me know).
But we will also take a look at this point and try to improve it.

About the Linux-build - we finished the wrapper for MaryTTS and it works great! This means, soon there will be the possibility to use our asset with the same voices on all Unity platforms (as long as there is an Internet connection available)!

The last questions: "Enforce 32bit voices" is only necessary if you have some needed (3rd party) 32bit-voices on a 64bit Windows system. This setting has no effect on a 32bit Windows system - it uses automatically the correct voices.

I hope this answers your questions, please let me know.


So long,
Stefan

 

Some voice names with accents don't seem to play nice with RTVoice. For instance IVONA 2 (Spanish (US)). "Penélope", the "é" gets the "?" treatment. Any workarounds while still using names?

Thanks

 

Hi

Can you please send me an email with the following details:

• some screenshots showing the problem
  
• logs (if exists)
• OS-version
• Unity-version
• RTVoice-version (PRO/Standard)
  
• invoice number

Thank you!


Cheers
Stefan

 

RE: Penélope voice. We had the same problem, on Windows. We ended up editing the registry entry to remove the accented character. Permanent_AfterGlow are you using Windows, Android or what? (We would like to get Ivona installed on Android but are having no luck getting a response from them...).

 

Stefan, we are getting reports from our QA on Android that sometimes Speaker.Voices returns null. It seems it's only on certain devices. Of course it doesn't happen on mine. It feels like a timing issue. Any thoughts?

 

Hi rgonsalv

We are aware of this problem and fixed it already.
If you send me the invoice I could provide you the new version in a few days (we currently testing the next release).

Cheers
Stefan

 

Thanks rgonsalv for the workaround!

We're also aware of this problem but we couldn't fix it - we're still trying to find a "proper" way for the future, but I think this should help you further.

 

Yes sorry about my quick last post, I was in a hurry. Yes, I am using Windows 10 (64bit), Unity 5.5.0f3 Personal (64bit). Here is a small glimpse of what the Console outputs given a Debug.Log on items from Speaker.Voices:

Voice {Name='IVONA 2 Pen�lope', Description='IVONA 2 Pen�lope - Spanish (US) female voice [22kHz]', Gender='Female', Age='Adult', Culture='es-US'}
UnityEngine.Debug:Log(Object)
​

Quick glance at IVONA's site with other potentially problematic voices include:

Icelandic "Dóra"
Portuguese, Brazilian "Vitória"
French "Céline"
Portuguese "Inês"​

I otherwise have no issues with e-acute used as string or any other Unicode characters in VS or Unity. Thanks for the registry hack, I'll try that.

 

Hello again

I think we found a solution. Please send me an email with the invoice and I send you the fix.


Cheers
Stefa

 

Hello Stefan, I have a problem to change the actual speaker.
I'm using Dialog-System with Salsa and RT-Voice.
RT-Voice found 2 Voices on my system, the first one hedda a german voice the second one zira an english voice.
I'm not able to switch to the englisch voice.
I changed the preferences in the RT_Voice Actor Script to
language: en
name: zira
but i always get Hedda trying to speak englisch with german accent, quit funny but not useful in the long run.
Thank you for help with this one!