How to integrate google-speech + channel-vonage on botpress

Ask the Community
1

Hi everyone!

How do I integrate the feature to transcribe audio to text in botpress

Is it possible to send a spoken audio through the web channel and botpress will transcribe this audio to text?

I set up google-speech with channel-vonage but still can’t get it to work.

tks

2

Hi @danilocesar.lima ,

What version are you using?

3

Hi @bassam.tantawi, I used the image: botpress/server:v12_26_9 version.

Thanks

4

Hi @danilocesar.lima ,

This forum is for the Botpress Cloud version. The Botpress v12 version can be found here: Discussions · botpress/botpress · GitHub

But here is simple steps to activate google-speech directly in Botpress, without the need for Vonage.

Bot-level Configuration

Enable the Google Speech module

  1. Go to the module admin page and unpack the google-speech module, if it was already unpacked then enable/activate the module

1258×906 84.7 KB

  1. Wait for the module to be enabled/activated

  2. Go back to your bot, and open the bot

  3. Go to the code editor tab, then right-click the google-speech.json and choose “Duplicate to current bot”

  4. 624×590 40.1 KB

  5. Set the following properties:

  6. Prerequisites:

* The google cloud privateKey file is generated from the steps applied on the “Requirements” section.
  1. Configuration file parameters:
* enabled: true
* clientEmail: You will find this value in the JSON file containing the service account keys.
* privateKey: You will find this value in the JSON file containing the service account keys.
* projectId (Optional): The project ID on which the Google APIs are enabled.
* confidenceThreshold: The confidence threshold used to discard text alternatives when using speech-to-text. Must be a floating number between 0 and 1
* maxAudioDuration: The maximal duration (in seconds) allowed for an audio file when using speech-to-text.Must be a number between 0 and 60
* languageMapping: The mapping between Botpress languages and languages in BCP-47 format recognized by Google.
* voiceSelection: The type of voice the text-to-speech audio file should have. One of: "MALE", "FEMALE", "NEUTRAL".

|759.0133094787598x209
|759.0133094787598x2091600×442 142 KB

  1. Restart Botpress Server to reload the configuration

Allow audio MIME type

  1. Go back to your bot, open the bot
  2. Go to the code editor tab, open botpress.config.json
  3. Scroll down to the allowedMimeTypes tag
  4. Then add “audio/webm”

947×338 30.7 KB

Allow the Microphone button

  • The last step is to activate the Microphone button by using the “enableVoiceComposer” flag, either through the initiation script or using the debugger

1600×699 132 KB

5

Module Limitations

Currently, channel-vonage and channel-web supports receiving and sending voice messages

For more about Vonage integration, refer to the Vonage documentation.

Speech-to-text

Currently, this feature only supports ogg/obus and mp3 (mpeg 1 layer 3) audio files. When the module receives a file in a format that is not supported, a warning will be displayed in the logs.

Text-to-speech

Only sends .mp3 files for the moment.

Appendix A - Links

Google-Speech on Botpress Github:

Web-Channel options:

Appendix B - Screenshots

536×545 16 KB
1600×138 12.6 KB
1600×867 150 KB
691×163 15 KB
562×649 29.7 KB
839×384 20.1 KB
1600×828 245 KB
1166×893 82.9 KB
1037×627 49.1 KB
1030×455 41.3 KB
942×885 74.2 KB

6

Hi @bassam.tantawi thank you for your support!

I have another question for you. Where can I configure “enableVoiceCompose: true” parameter
without being in debug mode?

7

What is the specific directory to change this configuration file?

8

What is the specific directory to change this configuration file?

9

Hi I don’t see the module admin page on Botpress Studio, I’m running 1.3.1 and the options I have in the sidebar are: Designer, Content, Chatbot Information

Any ideas where I can find this? Thanks!

10

When injecting the bot

11

@leolambert.sc Hi, sorry this is for the OSS, this integration can be done when the APIs layer is released. That should be somewhere around mid March

12