Speech to Text - yes again

Distro Discussion & Help
1

I’m hoping for a new answer, a yes answer.

Can I add a reliable speech to text option to me FreePBX distro?
Can anyone using this please speak up.

Thanks in advanced.

2

We open sourced a lot of our speech to text stuff from years ago but it isn’t really maintained. That said it could be used as a guide on how to accomplish things. Some of the cloud services by Amazon and IBM for speech recognition are amazing. At Astricon I watched 3 people doing real time(ish) translation from language A to language B. There was a few second delay but it was super cool stuff.

With FreePBX I don’t know of anyone using it beyond a hobby level with random hacks. I have played with this stuff as a general API and would like to play with some direct integrations as a personal project but with a full schedule I have to find time for other personal projects.

3

Thanks James for the quick response.
The Amazon stuff looks good. https://aws.amazon.com/polly/pricing/
So I’m sure someone has this connected to FreePBX.

so who has this working?

4

This is the most progress I’ve seen in the forum. Users on the thread appear to have implemented this approach. As James mentioned, this is not a clean integration.

5

Sorry for late reply. I only just saw this post.
I had previously tried the Polly implementation and found it worked “out of the box”

I was recently asked to build a system that took a text message and phone number via API, convert the message to “voice”, ring the number and play the message. I have a prototype engine working using Polly, and it works well. Working on a production system now.

The prototype is using Freepbx as the asterisk engine, though I did the Polly conversion outside Freepbx under the API.

Let me know if you want some specific info.

1 Like
6

Please share them… Tx

7

Please do share.

8

We also went ahead using the linked method to incorporate Polly as a TTS engine. Works very well using the TTS GUI. I wish that this could be a more standard integration (a module with a GUI), but since I lack the skills to fix this, I don’t have room to complain.

9

Ok I will try and document a few key steps as I build this production server.

1 Like
10

We are using http://nexiwave.com/ for our voicemail. It works about as good as google’s voice->text on a phone.

11

I have posted a new example under the title Text to Speech - Revisited