72.7.2.11 Google Speech Recognition

By relying on Etere Media Asset Management is possible to transcript video and audio files into Subtitles Text, simply by executing an Etere T-Workflow against any wanted asset.

The system will transcript everything and assign text to timecode, so the operators would eventually find all the subtitles automatically populated while opening the given asset with Etere Subtitles Tools.

In order to start using this feature we may need to perform some procedures first.

1. update Etere to the required version;
2. install Etere Service Media Manager and Etere Speech to Text modules;

   google_speech_install
   

3. obtain Google Service Account JSON file(Etere giude on this topic is available on next paragraph > 78.7.2.12 );
4. insert the given key into:

   Etere Configuration Editor > Google Speech Recognition > Service Account File

5. set the Audio Channel to use

   NOTE: This option is not mandatory for the given feature.

   google_speech_settings
   

6. select the Language we want to use and the related Subtitles Language as well;

   subtitles_bcp47

   language_bcp47
   

7. create Google Speech to Text Action in:

   Etere Configuration Editor > T-Workflow > Actions > Google Speech Recognition

   and fullfill the required parameter:

   • Name:
   • Language:
   • Service Account File: 

     create_google_action

     google_action_done
     

8. once the action has been created, we can start creating our Etere T-Workflow

   speech2text_workflow
   

9. upon completing the given workflow, if the operation went successfully we can now open:

   Etere media Manager > Subtitles Tools

   recall the given asset and check if the text has been converted to Subtitles.

   subtitles_tool_speech_2_text

If the station have no license for the Google Speech Recognition integration, a message would eventually appear informing that the system is not able to login.

google_speech_to_text_missing_license

To use V2 as specified in chapter 59.3.15.3

In Subtitle Tools, the Recognizer Name to use is read from the Etere Configurations. If it is empty,
the V1 recognizer will use used. This maintains backward compatibility

Items needed to use the function:
1) Service Account file name (file json)

2) Recognizer Name, explained in the configuration chapter

3) Audio file, linked to the asset and with the codec set to audio enabled

The result will be displayed as in the example (Speech_Recognition_V2)