Search Unity

  1. Unity 2019.1 is now released.
    Dismiss Notice

[RELEASED]Unity3D Google Cloud Speech Recognition [VR\AR\Mobile\Desktop]

Discussion in 'Assets and Asset Store' started by FrostweepGames, Aug 6, 2015.

  1. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hi,

    are you use latest plugin version 2.1? From version 2.0 we are use latest Google Cloud Speech API https://cloud.google.com/speech/

    Quota:
    http://dl3.joxi.net/drive/2017/01/13/0017/3644/1146428/28/4e5384e5bb.jpg
    Pricing:
    http://dl4.joxi.net/drive/2017/01/13/0017/3644/1146428/28/f790197859.jpg

    Please use latest API dont forget to configure Billing in your google account

    Hope helped you.

    Best
     
    Freddy888 likes this.
  2. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    if your montly usage < 60 min your Speech API is free
     
    Freddy888 likes this.
  3. Freddy888

    Freddy888

    Joined:
    Sep 13, 2014
    Posts:
    164
    Thanks.

    So I can forget about that Chromium Speech API now then ?
     
  4. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    For now we are not supporting Chromium backend.. because we had some problems with clients who wants latest Google Cloud Speech API..
    But chromium is still free for use
     
  5. Freddy888

    Freddy888

    Joined:
    Sep 13, 2014
    Posts:
    164
    Thanks for the quick replies :)

    I have it working now. I reduced the "Record Length" setting down to 2 to get quicker results if anyone else needs to try that.

    Seems to be working well in normal mode. Setting the Language to my locale seemed to help with errors.

    I'll wait until the next version to see how Runtime Detection can be improved.

    Keep up the good work :)
     
  6. Eltayeb

    Eltayeb

    Joined:
    Jan 24, 2014
    Posts:
    15
    so i found out that i need to buy speech api from google in order to run my application which i cant afford right now, is ur other plugin
    Yandex Cloud Speech Recognition work for free or i have to buy something as well ?
     
    Last edited: Jan 14, 2017
  7. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,
    Here is detailed info about use Yandex Speech Kit https://tech.yandex.com/speechkit/cloud/doc/intro/faq/concepts/about-docpage/#q4_30
    Our asset based on SpeechKit Cloud for cross-platform uses.
    http://dl3.joxi.net/drive/2017/01/15/0019/1302/1250582/82/955e5de983.jpg
    Actually "free for use" services for speech recognition do not exists..
    You can make neural based system on your servers for speech recognition, this is a not good idea and need big servers but it for free for use, only pay for servers.

    Google price plans is good and you can get 60minutes per month for free..
     
  8. Eltayeb

    Eltayeb

    Joined:
    Jan 24, 2014
    Posts:
    15
    so i would need an api key from them in order to be able to use it right ?

    .
    do you know a tutorial on how to do that?
     
  9. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
  10. Eltayeb

    Eltayeb

    Joined:
    Jan 24, 2014
    Posts:
    15
    sry i meant for Yandex speechkit do i need an api key or if i get ur plugin it'll work without one?
     
  11. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    you need your own api key.

    we cant pay for requests instead our clients
     
    Eltayeb likes this.
  12. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello again,
    I asked Google team about time for Speech Recognition and they said that:

    "With regards to your question, the upload time of a file to Google Service varies depending on which country you are in and what network provider you have. Then since you are using a synchronous API recognition, the time of the recognition will take approximately the same with the length of the audio file or more."

    Hope helped you.
    Best
     
  13. Eltayeb

    Eltayeb

    Joined:
    Jan 24, 2014
    Posts:
    15
    hello,
    can you tell me if i can detect 2 languages at the same time??

    best regards
     
  14. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    hello,

    hm, good question.. but I guess no..

    need to ask the google team

    hope helped you,

    best
     
  15. chihchanglin

    chihchanglin

    Joined:
    Jul 24, 2015
    Posts:
    3
    Hi, According to the FAQ above, I speech a very short word like "hello". I think the length of this audio file should around 1 second, but the time of the recognition from Google Cloud takes 5 seconds or more.

    any suggestion?

    best regards
     
  16. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hi,


    looks strange but as google team said "time of the recognition will take approximately the same with the length of the audio file or more"..

    On client side (in unity) we have fastest api for sync recognition.. but "runtime speech detection" code need to improve.. also we guess need fast internet connection (not 2g:D )
    also we research more better audio codec for speech recognition on google side

    best
     
  17. chihchanglin

    chihchanglin

    Joined:
    Jul 24, 2015
    Posts:
    3
    Hi
    I'm using WiFi and the speed is 16Mbps/8Mbps.
    Is there has any way I can change the code to improve the time of the recognition?

    best regards
     
  18. Etechanint

    Etechanint

    Joined:
    Mar 4, 2015
    Posts:
    1
    Hey

    I've seen in the details it includes +88 languages.
    I'm quite positive Arabic is included, but I'd like to know for sure about this.
    It includes Arabic language ?
    Also can I compare an output text from the speech with a text inside my project ?

    Like a reading test.

    Thank you
     
    Last edited: Feb 26, 2017
  19. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,

    yes, Arabic language is supported by google.

    also you can compare recognition output with your text

    Best
     
  20. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    If you use runtime speech detection - yes, you can improve code. but if you use start\stop record - no, because all works fine and time for the recognition depends from google cloud and internet connection
     
  21. fafase

    fafase

    Joined:
    Jul 3, 2012
    Posts:
    156
    Hei,

    Just bought the asset but I am confused as to the Google Cloud thing. Will I be charged extra by Google for using the plugin?

    Regards
     
  22. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,

    Yes, google Service have a pricing plans:

    and quote from asset description: http://dl3.joxi.net/drive/2017/03/13/0017/3644/1146428/28/2c87039914.jpg

    Hope helped you,
    Best Regards
     
  23. fafase

    fafase

    Joined:
    Jul 3, 2012
    Posts:
    156
    Thanks.
    You should replace that line with "The plugin does not cover the cost of Google Cloud Service" or something of the kind explicitly mentioning this is not the whole cost for using the plugin.

    Is there any offline usage by the way?

    EDIT: I managed to run the app on device. It seems to require the user to start and stop the recognition manually. Could it be done automatically when the user stop speaking like it does in Windows Speech? Maybe there is an example in the package that does that.

    Also, it is fairly slow despite having a quite fast connection just to get "Hello".
     
    Last edited: Mar 13, 2017
  24. fafase

    fafase

    Joined:
    Jul 3, 2012
    Posts:
    156
    Just bumping this one.

    Is there any offline usage by the way?

    I managed to run the app on device. It seems to require the user to start and stop the recognition manually. Could it be done automatically when the user stop speaking like it does in Windows Speech? Maybe there is an example in the package that does that.
     
  25. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,

    Yes, you are right.. We need to add this line..(will be added in new update of our asset).

    Also, our asset doesnt supported offline mode because he provide only communications between Google server and unity application.

    Our asset have a runtime speech detection(current version will be deprecated, new version available in beta\alpha mode in asset settings)

    Sorry for the long delay..

    Best Regards
     
  26. ninjanosui

    ninjanosui

    Joined:
    Jul 12, 2013
    Posts:
    27
    I also get : Speech Recognition failed with error: Process Parse Response failed with param: {}
    on Android.. In the editor it works fine.. Any idea what is causing this?
     
  27. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,

    Sorry for the long reply..

    I guess this is the problem with recorded sound.. any words not detected.

    Also r u use the latest version of our asset?

    Best
     
  28. wtf3d

    wtf3d

    Joined:
    Dec 14, 2014
    Posts:
    3
    Hello!
    What is this error for?
    Unity 2017.1.0b2

    ArgumentException: Cannot set Request Header Content-Length - name contains illegal characters or is not user-overridable
     
    Last edited: Apr 28, 2017
  29. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,

    I guess Unity have updated core api for WWW requests.. In future update we will update web requests.

    Also this can be bug on engine side because it 'Beta'.. we will check it

    Thank you.

    Regards
     
  30. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello guys! We are announce new version of Google Cloud Speech Recognition 3.0 .

    This is a big update what include new features and fixes:
    * New Plugin Core
    - New Networking
    - Multi-threading
    - Media Manager
    ~ More optimized AudioClip convert API
    ~ Updated WebGL microphone API
    ~ Other features
    - New low level api
    - Runtime Speech Detection Manager
    * Upgraded Speech Recognition API from the version 1beta1 into the version 1
    * Supporting of the sync and async recognition API
    * Prepared project to implement streaming speech recognition (only on Dekstop: Win\Mac\Linux - not on the mobile platforms) - Require the .NET Core and .NET Framework 4.5+ in Unity Build settings.
    * And much, much more

    Warning:

    Dont forget upgrade* your project to version 3.0 before 31 July 2017. Google will close supporting of old API version.


    * new version of our plugin will be available soon in Asset Store.

    Thanks for the support us and good luck!

    Best Regards,
    Artem Shyriaiev
     
  31. Ascensi

    Ascensi

    Joined:
    Sep 7, 2013
    Posts:
    567
    Does this mean/allow players not have to prerecord their voice to get detection to work?
     
  32. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Yes, you are absolutely right;)
     
    Ascensi likes this.
  33. AteoMarcel

    AteoMarcel

    Joined:
    Jan 5, 2017
    Posts:
    17
    Do you have some ETA on when your update (including streaming) goes online?
     
  34. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hi, I've sent you email.

    "I guess in version 3.2 we will add streaming speech recognition..
    Also in version 3.0 we improved runtime voice detection and decrease time for the working on client app.. yes that is not streaming but can help..
    actually we are working on streaming speech recognition but its still on alpha version - that is because Unity 2017.0.1 in beta also and we have some troubles with grpc in Unity"

    Also I guess our product is big and we need to make road map for understand when and what features we will integrate in our asset.
     
  35. Soshiyanet

    Soshiyanet

    Joined:
    Feb 6, 2017
    Posts:
    1
    Hello FrostweepGame! Thanks for your wonderful job. I cannot wait to see you add streaming speech recognition. MY goal is:
    Talk through microphone in any of those 88+ languages and see its text and also according to my speech a voice play and answer my question or whatever I said. I am using Photon Voice for voice chat.

    Is it compatible? Can you give me a quick workflow?

    Thanks my man
     
  36. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hi, thanks for the interest to our asset:)

    Our asset works like this:
    ~ Without runtime voice detection:
    - Start Record,
    - Detect when talking is started
    - Actually recording
    - Stop Record
    - Convert Audio to bytes
    - Generate Request with data
    - Send request and waiting the response
    - Deserialize response data
    - Other working with recognized text

    So you can work only with recognized text. But you can make your own speech synthesis module for play recognized text and of course, before - enable recording of the Photon voice chat and when speech are finished disable recording of the Photon voice chat.

    Good asset for the playing voice from text: https://forum.unity3d.com/threads/rt-voice-run-time-text-to-speech-solution.340046/

    I hope helped you!

    Best Regards,
    Artem Shyriaiev
     
  37. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
  38. jtesler

    jtesler

    Joined:
    Jul 19, 2017
    Posts:
    6
    I just noticed that in version 3.0, you have a field to enter the API key in GCSpeechRecognition, however you ignore it and use the hardcoded value from Constants. So everyone using the 3.0 library is using your demo key. The previous version did use the entered key.
     
  39. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,

    Actually everyone can change api key from constants directly in code or can add own api key in GCSpeechRecognition prefab in text field and enable Use API Key from prefab checkbox.

    Also in documentation we are wrote about that.

    Hope helped you.

    Thanks and Best Regards,

    Artem Shyriaiev
     
  40. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Guys, we are noticed about problem with ThreadManager.cs in WindowsStore platform(Hololens).. if you will have problems with it just delete this script.. we will fix that problem in the next update.

    Best
     
  41. jtesler

    jtesler

    Joined:
    Jul 19, 2017
    Posts:
    6
    Regarding the API key, the key in the GCSpeechRecognition prefab is ignored, even if the UseApiKey checkbox is checked. It only uses the value in Constants (which I changed).
     
  42. shmowlo

    shmowlo

    Joined:
    Aug 14, 2014
    Posts:
    1
    Hi Artem @FrostweepGames

    How are you managing to use the Google.Cloud.Speech API in Unity when it requires .NET 4.5? Have you wrapped your code somehow, or do you create your own headers/httpclients and send your requests direct?
     
  43. MHeru_Art07

    MHeru_Art07

    Joined:
    Nov 7, 2016
    Posts:
    2
    Hello, i want to ask.. is this plugin can process speech recognition in indonesian languange with a good result? If yes, can you show me the demo apk in indonesian language?

    Thankyou
     
  44. omreaney

    omreaney

    Joined:
    Jul 26, 2017
    Posts:
    1
    Hi @MHeru_Art07, this plugin is only a mere helper to access Google Cloud Speech API so you can check on this page if your language is available ( as I see it, it is ;) )
     
    FrostweepGames likes this.
  45. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hi, actually we use REST api for sync speech recognition that not require .net 4.5 and we use UnityWebRequest's.

    For Streaming speech recognition we use .net 4.6 on Unity 2017.1 (that is new version of our asset)

    Hope I helped you!

    Thanks and Best Regards,

    Artem Shyriaiev
     
  46. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hmm,

    I will check it. Thanks for the report!
     
  47. woo_Ose

    woo_Ose

    Joined:
    Dec 1, 2016
    Posts:
    9
    FrostweepGames is it support async/long recognition (5 min rocognition)???
    i want to ask first before buy it, if no maybe next update???
     
  48. FrostweepGames

    FrostweepGames

    Joined:
    Jan 2, 2015
    Posts:
    200
    Hello,

    in the version 3.0 - no.
    in the version 3.1 - will be in asset.

    update 3.1 will be available soon in asset store (September I guess.. also we hope to add the streaming speech recognition).

    Thanks
     
    woo_Ose likes this.
  49. woo_Ose

    woo_Ose

    Joined:
    Dec 1, 2016
    Posts:
    9
    oh thanks, i'll wait then. Hope it ready asap.
    really needed for my project tho
     
  50. TayoEXE

    TayoEXE

    Joined:
    Jul 27, 2017
    Posts:
    5
    FrostsweepGames, after purchasing your asset, I have been painstakingly trying to implement it according to the instructions given in the tutorial and videos, but we are now wondering if there is an issue with the latest build you have available? Particularly, we found that there were no actual Config objects created for line 79 of the GCSpeechRecognition.cs file

    _speechRecognitionManager.SetConfig(configs[currentConfigIndex]);

    We were wondering if this was a mistake, and if not, how do we fix this? We are in a hurry, and we have not been able to get this working. Thank you for any information. (And I tried sending an email, but I never received a reply, unfortunately.)