Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

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

Discussion in 'Assets and Asset Store' started by Stefan-Laubenberger, Jul 10, 2015.

  1. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi

    Are you using the latest version 2.9.8 from RT-Voice?
    Which is your current TTS system - Windows, MaryTTS or eSpeak?


    Cheers
    Stefan
     
  2. mengux

    mengux

    Joined:
    Jan 12, 2018
    Posts:
    7

    I am using RT-Voice PRO 2.8.5
    TTS system is Mac? (I am not using MaryTTS or eSpeak)

    The pitch problem also happened when I am using prefab like SpeechText & TextFileSpeaker, changing pitch setting in inspector doesn't actually change the pitch, but function like rate works.
     
  3. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    I'm a bit confused. :( macOS doesn't support pitch nor SSML - at least on my machines...
    Where did you test the SSML (prosody) text, on your Mac and inside Unity?
     
  4. mengux

    mengux

    Joined:
    Jan 12, 2018
    Posts:
    7
    Wow, then it becomes surprise. It did work on my Mac inside Unity.
     
  5. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Sorry I don't get it... My Macs have no support for SSML. Are you sure you didn't use MaryTTS by accident? Can you please provide me with more information about your system and your test setup? Which voice have you tested?
     
  6. BookMak

    BookMak

    Joined:
    Sep 2, 2015
    Posts:
    6
    Hi Does RT-Voice work on Hololens?
     
  7. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi

    It's a simple "yes!" ;)


    Cheers
    Stefan
     
  8. BookMak

    BookMak

    Joined:
    Sep 2, 2015
    Posts:
    6
    Hi, Whats the differene between RT-Voice & RT-Voice Pro?
     
  9. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    The PRO has the source code, works with WSA (UWP) and receives bugfixes outside of our normal deployment cycle (= normally once per month).
     
  10. mengux

    mengux

    Joined:
    Jan 12, 2018
    Posts:
    7
    Seems find the reason, the voice I use is from cereproc, it is not default voice. Although it is not marry tts, maybe it works the same.

    Another thing is when I found Mac OS won't work, I changed to Window, we have several windows machine, one works, another two didn't work when change the pitch, is there any common conflict or mistakes may result right? Sorry for such a general question, the source code is pretty big, cannot post it here.
     
  11. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Some Windows voices don't support SSML (especially pre Windows 10) which is needed to use the pitch. There is no native pitch in Windows, only via SSML.
    But, you could use the pitch from the AudioSource instead ;-)
     
  12. icarus1564

    icarus1564

    Joined:
    May 26, 2018
    Posts:
    5
    Good evening... I'm a newbie working with RT-Voice, and I've scoured this thread but I haven't found anyone posting about my particular issue. I just purchased and downloaded RT-Voice, and it is amazing -- thank you for your great work! -- but it seems that it can not play any voices other than "en-US" culture. Here's the dump from the debug script you posted a while back:

    "
    Voice {Name='IVONA 2 Amy', Description='IVONA 2 Amy - British English female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='en-GB'}
    Voice {Name='IVONA 2 Céline', Description='IVONA 2 Céline - French female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='fr-FR'}
    Voice {Name='IVONA 2 Dóra', Description='IVONA 2 Dóra - Icelandic female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='is-IS'}
    Voice {Name='IVONA 2 Emma', Description='IVONA 2 Emma - British English female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='en-GB'}
    Voice {Name='IVONA 2 Gwyneth', Description='IVONA 2 Gwyneth - Welsh English female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='en-GB'}
    Voice {Name='IVONA 2 Ivy', Description='IVONA 2 Ivy - US English female voice [22kHz]', Gender='FEMALE', Age='NotSet', Culture='en-US'}
    Voice {Name='IVONA 2 Jennifer', Description='IVONA 2 Jennifer - US English female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='en-US'}
    Voice {Name='IVONA 2 Maja', Description='IVONA 2 Maja - Polish female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='pl-PL'}
    Voice {Name='IVONA 2 Nicole', Description='IVONA 2 Nicole - Australian English female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='en-AU'}
    Voice {Name='IVONA 2 Raveena', Description='IVONA 2 Raveena - Indian English female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='en-IN'}
    Voice {Name='IVONA 2 Salli', Description='IVONA 2 Salli - US English female voice [22kHz]', Gender='FEMALE', Age='Adult', Culture='en-US'}
    Voice {Name='Microsoft David Desktop', Description='Microsoft David Desktop - English (United States)', Gender='MALE', Age='Adult', Culture='en-US'}
    Voice {Name='Microsoft Hazel Desktop', Description='Microsoft Hazel Desktop - English (Great Britain)', Gender='FEMALE', Age='Adult', Culture='en-GB'}
    Voice {Name='Microsoft Zira Desktop', Description='Microsoft Zira Desktop - English (United States)', Gender='FEMALE', Age='Adult', Culture='en-US'}"

    So you see I have several en-GB and other voices. When I attempt to select anything other than "en-US" voice, I get the windows default voice.

    I've played around with the voice demo a bit, trying to see if I could force it to play "IVONA 2 Amy" for example, but it would only play the default voice. All these voices are currently working on my windows box.

    Thank you for your help (and for your excellent asset :) )!
     
  13. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi

    If a voice doesn't support SSML, the Windows TTS system will always fallback to the default voice... RTV uses SSML per default (for the pitch), but unfortunately, we can't detect which voices support SSML.
    The only thing you can do is enable "Auto Clear Tags" on the RTVoice component (which subsequently disables SSML):
    upload_2018-7-2_9-22-23.png

    This is the way to do it, BUT I just discovered a bug in the latest version which prevents this approach from working... :(
    Do you have the PRO version? If so, I could send you the fix immediately...


    Cheers
    Stefan
     
    icarus1564 likes this.
  14. icarus1564

    icarus1564

    Joined:
    May 26, 2018
    Posts:
    5
    Yes I do have the pro version... a fix would be very helpful :)

    I'm also playing around with multiple NPCs in the same scene... any pointers or examples for having each speak with a separate voice?

    Thank you again for your excellent work and fast response.
     
  15. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Great, please send me your invoice and I can give you the fixed version.
     
  16. icarus1564

    icarus1564

    Joined:
    May 26, 2018
    Posts:
    5
    Sent. Thank you!
     
  17. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Fix also sent ;)
     
  18. icarus1564

    icarus1564

    Joined:
    May 26, 2018
    Posts:
    5
    Awesome, thank you!

    I am trying to set up a scene with several NPCs (hence my voices request) in VR. I've tried several approaches to getting the 3D audio aspects to work properly, including utilizing the RTVoice wrapper directly and pasing in the individual gameObject AudioSource, but the audio still seems to come from everywhere and not from the individuals who are speaking.

    I've also implemented my own lipsync using Genesis3 and Genesis8 morphs (those are my base models) -- everything is working great subscribing to the OnSpeakCurrentVismeme event (I had to filter the speaker based on the voiceName I chose for a specific NPC, otherwise ALL characters recieved ALL vismeme events...).

    If I can solve the voices issue in my original post AND get the 3D audio to function properly I will be off to the races with an awesome framework, thanks to your fine work.

    Any tips or pointers on 3D sound, RT-Voice, and VR?
     
  19. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Have you set "Spatial Blend" to 1 on the AudioSource? ;)

    About the visemes: you don't have to filter it by voice name. Every "Speak"-call returns an UID which is also part of the wrapper. Therefore, if you listen to the callbacks, simply distinguish the different origins by UID.
     
  20. icarus1564

    icarus1564

    Joined:
    May 26, 2018
    Posts:
    5
    Makes sense... and yes, I have played around with pretty much every setting on the AudioSource, starting with the Spatial Blend. With and without "spatialize" selected. Also tried various combinations with the Oculus Rift ONSP Audio scripts (with scripts, without scripts, etc). Got a dumb primitive to act like a boom box looping an audio sound...

    None of it worked with the RT-Voice generated sounds. The output of the voices are all uniform volume no matter where the audio receiver is in the room, and the audio always sounds like it's front-and-center.

    It might make a difference that I'm using "SpeakNative"... there's basically no data for it to chew on before playing. Do you think that could be it?

    Have other folks managed to make RT-Voice with multiple audiosources work in VR? There's probably something simple I've missed... but everything else seems to work.

    ----
    [EDIT] Your updated code worked! I now have access to all my IVONA 2 . Thank you.
     
    Last edited: Jul 4, 2018
  21. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    SpeakNative won't work. You need the "Speak"-method and give it an AudioSource as parameter. For more, please see the API:
    https://www.crosstales.com/media/da...ass_crosstales_1_1_r_t_voice_1_1_speaker.html
     
  22. iVizUnity

    iVizUnity

    Joined:
    May 23, 2018
    Posts:
    3
    We use and love your product - have it working with our own AI front end that can have a conversation with a user - we pass the current genreated AI sentence to RTVoice - then through Salsa - which then creates a naturally speaking and gesture 3D avatar in Unity. We use different voices that RTVoice can take and as well we get timing out of RTVoice for each worrd spoken so we can fire animation ( gestures) timed with the words.

    Great Product.

    Question: What is the best sounding voice ( to buy - we use ceroproc Adam now) that works well with your system - one that including strong ssml prosody (natural speaking voice commands) with it?

    -steve
    Prof, Simon Fraser University
    ivizlab.sfu.ca
     
  23. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi Steve

    Thank you for the kind words!

    Personally, I think the voices from IVONA are the best:
    https://www.ivona.com/us/about-us/voice-portfolio/

    I like Brian and Amy, but the others are also great.


    Cereproc is also very good:
    https://www.cereproc.com/storesapi

    There I like Giles and Lauren.

    In the end, it's only a matter of taste... ;)


    Cheers
    Stefan
     
  24. mengux

    mengux

    Joined:
    Jan 12, 2018
    Posts:
    7
    Hi Stefan,

    Just want to confirm, does RTVoice support pitch change of Cereproc Voice? (Windows system.)
    Test pitch change by adding Prosody; using different prefab (textToSpeech, etc.) Cereproc voice's pitch seems cannot be modified, at the same setting, windows default voice works.

    Thanks in advance.
     
  25. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hello again

    I don't have any Cereproc voices, but it's possible since not all voices support the complete SSML-specification...
    As an alternative, you could use the pitch of the AudioSource instead.
     
  26. Arkade

    Arkade

    Joined:
    Oct 11, 2012
    Posts:
    654
    Hi
    I've been using RT-Voice (not Pro, I'm afraid) successfully in my WIP Android VR kids game (for users who cannot read). It's great, thanks!
    Just decided to do a build with IL2CPP but got failure that seems caused by RT-Voice. I've attached what looks like the relevant excerpt of the error log. An example snippet (for future searchers):

    F:/Users/User/tmp/VRToy56Direct/VRToy56Direct/Library/il2cpp_android_armeabi-v7a/il2cpp_cache/D0DE46227DCB7D43C1C4A0D10ED961EB.o: In function `VoiceProviderIOS_GetVoices_m3868273113':
    F:\Users\User\tmp\VRToy56Direct\VRToy56Direct\Temp\StagingArea\Il2Cpp\il2cppOutput/Bulk_RTVoice_0.cpp:17650: undefined reference to `GetVoices'
    F:\Users\User\tmp\VRToy56Direct\VRToy56Direct\Temp\StagingArea\Il2Cpp\il2cppOutput/Bulk_RTVoice_0.cpp:17650: undefined reference to `GetVoices'
    F:/Users/User/tmp/VRToy56Direct/VRToy56Direct/Library/il2cpp_android_armeabi-v7a/il2cpp_cache/D0DE46227DCB7D43C1C4A0D10ED961EB.o: In function `VoiceProviderIOS_Stop_m3308791736':
    F:\Users\User\tmp\VRToy56Direct\VRToy56Direct\Temp\StagingArea\Il2Cpp\il2cppOutput/Bulk_RTVoice_0.cpp:17640: undefined reference to `Stop'
    F:/Users/User/tmp/VRToy56Direct/VRToy56Direct/Library/il2cpp_android_armeabi-v7a/il2cpp_cache/D0DE46227DCB7D43C1C4A0D10ED961EB.o: In function `VoiceProviderIOS_Speak_m2841170132':
    F:\Users\User\tmp\VRToy56Direct\VRToy56Direct\Temp\StagingArea\Il2Cpp\il2cppOutput/Bulk_RTVoice_0.cpp:17668: undefined reference to `Speak'
    F:/Users/User/tmp/VRToy56Direct/VRToy56Direct/Library/il2cpp_android_armeabi-v7a/il2cpp_cache/D0DE46227DCB7D43C1C4A0D10ED961EB.o: In function `VoiceProviderIOS_Stop_m3308791736':
    F:\Users\User\tmp\VRToy56Direct\VRToy56Direct\Temp\StagingArea\Il2Cpp\il2cppOutput/Bulk_RTVoice_0.cpp:17640: undefined reference to `Stop'
    clang++.exe: error: linker command failed with exit code 1 (use -v to see invocation)

    Any thoughts on how to resolve this would be appreciated! Thanks :)
     

    Attached Files:

  27. mengux

    mengux

    Joined:
    Jan 12, 2018
    Posts:
    7
    Thanks for replying:)
     
  28. mengux

    mengux

    Joined:
    Jan 12, 2018
    Posts:
    7
    The only issue about changing Audio Source is it seems need to generate audio files first, I hope the speaking is real time. Let me exploring more, appreciate for keep answering my questions!
     
  29. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi

    Yes you are right, IL2CPP under Android doesn't work with the standard version.
    Please consider upgrading to the PRO version.


    Cheers
    Stefan
     
    Last edited: Jul 8, 2018
  30. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    This is no problem, The RTV "Speak"-method generates the audio file and adds it to the AudioSource where you can modify the pitch. This is very fast, no worries.
     
  31. Arkade

    Arkade

    Joined:
    Oct 11, 2012
    Posts:
    654
    oh no, when did that change? :( I deliberated which to buy when they were last on sale for almost a week! I chose based on the "Standard Edition" stating I needed Pro only if I "need the source code or build for WSA (UWP)". Should this instead read "build for IL2CPP" ?
     
  32. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Unfortunately, this is the case. I wasn't aware of this fact either - I thought it worked with pre-Unity 2017, but never tested it again with 2017+.
    IL2CPP under the standard version still works with WebGL and iOS.
    I apologize for the inconvenience and I changed the store description.
     
    Arkade likes this.
  33. Deleted User

    Deleted User

    Guest

    Hi!

    I appreciate this asset, it's been really handy for my team's particular game.

    I have a issue that I need help resolving; we are currently converting the English dialogue we have to Chinese. While experimenting with Windows Chinese voice packs, we found that supplying Chinese characters gave the best sounding dialogues outside of Unity.

    However, when trying to supply the same Chinese characters into RT Voice, it gives the error: "Sound could not be played. FMOD Error: An invalid parameter was passed to this function.". Any insights? I definitely am passing in a string into Speaker.Speak(...), as I'm debugging into console and it shows the Chinese characters perfectly.

    Thanks,
    Andy
     
  34. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi

    FMOD is part of Unity, so I think the error has another reason...

    Can you please tell me your Unity, RTV and OS version?
    I also like to see the text that causes the problem to investigate it on my side.

    Thank you!


    Cheer
    Stefan
     
  35. Deleted User

    Deleted User

    Guest

    Sure!

    Unity 2018.1.2f1
    RT 2.9.7
    Windows 10 Home

    The text that I was trying to supply is:

    https://translate.google.com/#auto/zh-CN/我是你的对手,测验女王!你要走了

    Sorry for the mess, but Unity forum won't let me post with Chinese characters? It considers it as spam.

    All English translation works fine when I switch the language back to English. If there isn't any Chinese characters, it defaults to speaking English using the Windows Microsoft Tracy (zh - HK, I debugged it and it says that the voice it's going to use), although it uses David's voice for some reason.

     
  36. andythai

    andythai

    Joined:
    Aug 3, 2017
    Posts:
    9
    ^ I'm Dyxanh.

    Just wanted to post an update; we updated our RT Voice version to 2.9.8, the problem persists with the same error.
     
  37. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    It took me some time to analyse the problem and I found the reason for this error. The Windows TTS system generates an empty audio file because of SSML. It seems your Chinese voice can't use SSML and therefore a non-Chinese voice (David?) is used, which can't speak the Chinese characters - this leads to the FMOD error...

    However, please make sure to disable "Auto Clear Tags" as described above:
    https://forum.unity.com/threads/rt-...o-speech-solution.340046/page-16#post-3550780

    But you also need the latest beta version of 2.9.9... :( Do you have the PRO version?
     
  38. Deleted User

    Deleted User

    Guest

    I see, I'll disable the Auto Clear Tags as you instruct. However, would that resolve this issue? What steps can I take otherwise?

    No, I purchased the standard version. I updated via Unity's notification regarding new version. Where can I download the beta version?

    edit: Looking at that post and what I had in editor, it was already disabled. I enabled it instead. Hoping to hear back from you soon regarding the fix!
     
    Last edited by a moderator: Jul 11, 2018
  39. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    I'm not 100% sure that the fix will solve the problem, but I think it's the most obvious reason.

    We are currently finishing version 2.9.9 but it's not ready for the official UAS release.
    This is one of the main advantages from PRO: hotfixes outside the release cycles. This enables us to have a feedback-loop with our customers and makes sure that their expectations are met.

    Therefore, the only two possibilities right now are to wait for the official release (planned for the end of next week) or upgrade to PRO to gain access to the latest version.
     
    Last edited: Jul 11, 2018
  40. hkfroggy

    hkfroggy

    Joined:
    Oct 16, 2012
    Posts:
    2
    Hi ,

    I have been using RT-voice and SALSA for 2 years, work great till recently upgrade for a new project.

    I have done a dry test on just the following plugin on a new project but still fail

    here is my test env on window unity 2018.2.0f2 , build for window

    updated to latest in asset store
    + RT-voice pro
    + SALSA with random eyes

    + a new project and new scene
    + import RT-Voice pro
    + import SALSA
    + unpack the SALSA pack in RT-Voice pro 3rd party folder

    run SALSA-2d - click the talk but and nothing happen , I can see this message in debug mode
    onSpeakAudioGenerationComplete : Wrapper {....}

    run SALSA_Bots
    I can hear the TTS voice from both of them , but the lips not moving

    run SALSA_demo
    voice is out, but Lips not sync and move

    run SALSA_Demo-ethan
    no voice

    hope this help you troubleshoot the issue.

    Cheers,

    - Richard
     
  41. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi Richard

    The version of SALSA we use for our demo-packages is quite old and leads to the problems you experienced.
    We will fix this asap.


    Cheers
    Stefan
     
  42. Alex_Brazil

    Alex_Brazil

    Joined:
    Sep 2, 2014
    Posts:
    6
    Hello,
    I purchased the RT-Voice asset recently and initially want to congratulate you on the great work. I would like a tip to find a solution to what I need to implement.
    The texts I am converting to speak are large and so a delay occurs when the audio starts to play for the user, causing the user to think that the application has crashed. I went to help with a method that would return the final moment of converting the audio text so that I could put a "Wait ..." message, but I did not succeed.
    Can you help me?
    If you have an example I thank you even more.
     
  43. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi

    Sure, there are multiple callbacks available, but the most two obvious are "OnSpeakAudioGenerationStart" and "OnSpeakAudioGenerationComplete".
    I attached an example, but unfortunately, RT-Voice has a little bug and "OnSpeakAudioGenerationComplete" is only called AFTER the speech is finished. Therefore, change the lines 10&11 and 18&19 to get the event as soon as the speech starts.
    If you have the PRO version, send me an email with the invoice and I can give you the latest version with the fix.

    For more, please see the API:
    https://www.crosstales.com/media/da...ass_crosstales_1_1_r_t_voice_1_1_speaker.html

    Or documentation, chapter 5:
    https://www.crosstales.com/media/data/assets/rtvoice/RTVoice-doc.pdf

    I hope this helps you further, please let me know.


    Cheers
    Stefan
     

    Attached Files:

  44. Alex_Brazil

    Alex_Brazil

    Joined:
    Sep 2, 2014
    Posts:
    6
    Hello
    Thanks for the help, solved part of my problem.
    I could not get the SpeakAudioGenerationComplete event when speech starts.
    I do not have the PRO version.
    Has anyone predicted that repairing fix for non PRO users?
     
  45. Alex_Brazil

    Alex_Brazil

    Joined:
    Sep 2, 2014
    Posts:
    6
    Hello again,

    After another attempt I was 100% successful. Appreciate.
     
    Stefan-Laubenberger likes this.
  46. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    We will release the fix in about 2 weeks, but I'm happy that the workaround is fine.

    If you have any further questions, don't hesitate to ask!
     
  47. hkfroggy

    hkfroggy

    Joined:
    Oct 16, 2012
    Posts:
    2
    Thanks Stefan,

    do you know when will it ready to be release ? we are waiting the patch to move on, or we will have to move back to older versions soon.

    Regards,
     
  48. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    We will release it in the next two weeks.
    However, this should not hold you back since RTV and SALSA will work together anyway. If you need help to achieve this, send me an email.
     
  49. FS9606

    FS9606

    Joined:
    Mar 12, 2015
    Posts:
    21
    Hi there. Great product. What do you recommend, to include paralanguage, within this tool? This would be the inclusion of nonverbal words and sounds. For example, umm, words that people speak between other words. There seems to be no way to include umm, clicks, clearing throat, sniff, ahh, mmm, hmmm and other paralanguage items that are part of everyday conversation. Do you have any thoughts or suggestions on how to include this?
     
  50. Stefan-Laubenberger

    Stefan-Laubenberger

    Joined:
    May 25, 2014
    Posts:
    1,975
    Hi

    You can use the SSML "audio"-tag like this:

    <audio src="https://www.crosstales.com/media/rtvoice/Miau.wav"/>is the sound of a cat.
    <audio src="https://www.crosstales.com/media/rtvoice/Ouch.wav"/>I hit my thumb!

    Unfortunately, this currently works only under Windows and WSA.

    For other platforms, you have to roll your own solution, e.g. using your own "trigger"-tags in the speech.


    Edit:

    If you are using Cereproc, there are additional possibilities with the "spurt"-tag:
    <spurt audio="g0001_004">cough</spurt>, excuse me, <spurt audio="g0001_018">err</spurt>, hello.

    http://wiki.hansonrobotics.com/w/Voice_Expression_Authoring_Notes
     
    Last edited: Aug 4, 2018