Google speech to text punctuation

[CKFIT]

We're testing out Google's Cloud Speech-to-text API and currently have it running in its most basic form. By default it doesn't punctuate the message at all, and it just spits out one silly-looking run on sentence. It does provide the ability to automatically add logical punction with the "enableAutomaticPunctuation": true command being added to the RecognitionConfig portion of the request. From what I can gather (and I certainly could be mistaken), that is done on the 3CX side, but I can't find any documentation on where that script resides within 3CX. Does anyone have any info on this?

Thanks in advance!

Windows Server 2019 VM
Enterprise 18.0 update 5 b418

 

[CKFIT]

@edossantos Seems to be the name I always see in rest of the forum posts pertaining to Google TTS. Hoping you can help me if you don't make it to this portion of the forums too often!
It looks like the CFD would use this for listening to menu prompts, etc, but has nothing to do with transcribing voicemails.

 

[CKFIT]

Bump. Is there anyone here that has experience with integrating this API? There doesn't seem to be much info anywhere except for the call tree, which (I think) doesn't apply here.

 

[edossantos]

Hello @CKFIT,

There is no way to customize how 3CX invokes the Google API for voicemail transcription. You can only configure what you have in the 3CX Console > Settings > Voicemail.

The only way to achieve this would be implementing this transcription yourself outside of 3CX. You can create a service that monitors new WAV files in the voicemail location:

• Windows: C:\ProgramData\3CX\Instance1\Data\Ivr\Voicemail\Extensions
• Linux: /var/lib/3cxpbx/Instance1/Data/Ivr/Voicemail/Extensions

Once you detect a WAV file, you can invoke the Google Speech API using your customized parameters, and then send an email with the transcription result.

 

[CKFIT]

Hello @CKFIT,

There is no way to customize how 3CX invokes the Google API for voicemail transcription. You can only configure what you have in the 3CX Console > Settings > Voicemail.

The only way to achieve this would be implementing this transcription yourself outside of 3CX. You can create a service that monitors new WAV files in the voicemail location:

• Windows: C:\ProgramData\3CX\Instance1\Data\Ivr\Voicemail\Extensions
• Linux: /var/lib/3cxpbx/Instance1/Data/Ivr/Voicemail/Extensions

Once you detect a WAV file, you can invoke the Google Speech API using your customized parameters, and then send an email with the transcription result.

Thank you for this info! Aside from the work required in setting up a separate email process, with our users being pretty well-versed in how they receive these emails to begin with we probably won't want to implement a new source of voicemail alert emails. I can just imagine how many calls we'll get asking "Why do I get TWO emails about voicemails now?!" or "Is this phishing?"

 

[edossantos]

Thank you for this info! Aside from the work required in setting up a separate email process, with our users being pretty well-versed in how they receive these emails to begin with we probably won't want to implement a new source of voicemail alert emails. I can just imagine how many calls we'll get asking "Why do I get TWO emails about voicemails now?!" or "Is this phishing?"

You can also disable voicemail emails from 3CX.

 

[CKFIT]

You can also disable voicemail emails from 3CX.

Yeah, but I was just using that as a way to pad my "I don't want to learn how to do everything else needed and then set it all up" stance.