Translate speech in a real time

Overview

You can translate speech in a real time between call participants

This can be set per Connection or on the Trunk , by making a Call Flow with action Translate speech coming from VoIP Number / Translate speech coming to VoIP Number

Works with sampling rates: 8 000, 16 000, 24 000, 32 000, 48 000 Hz

Languages supported by real time speech translation 🗣️

Setup

You need to create a Call Flow that translates speech in one or both directions, Then, enable that Call Flow on Trunk or Connection. We will show how to do this in both cases

Translate speech to/from SIP Connection

1. Go to Call Flow Builder 👷🏼🔧 on your app.voipnumber.net/callflowbuilder or to Call Flows page at app.voipnumber.net/callflows and select +Add (or +New). Name your Call Flow and select if it's A leg or B leg

1. Place Translate speech coming from VoIP Number / Translate speech coming to VoIP Number actions on the Call Flow. This tells VoIP Number to start translating

1. Configure actions. Select languages

1. Select TTS and voice (we recommend to always use asynchronous TTS (gcp async streaming) which has slightly reduced support for languages but makes latency to first audio byte independent from the speech length)

1. Adjust audio levels

💡 If you need translated speech to be louder than original speech, attenuate original speech (rather than amplifying output - gcp async produces quite loud speech so amplifying it may cause distortion)

1. Select this Call Flow on Inbound settings of your Connection

✅ You now have created a Call Flow that starts speech translation in both directions on an inbound call to Connection To set translation on outbound calls from Connection you would create A leg Call Flow instead

In a typical scenario you only want speech to be translated once in a single direction, so you setup translation in a single direction, and on both Inbound and Outbound settings. Do this consistently across speech translation enabled Connections to avoid translating speech twice

Alternatively configure a Trunk. With a Trunk a Call Flow is independent of who is calling

Translate speech on a Trunk

As before, we will setup speech translation in both directions. We will use a Trunk to create a channel that accepts a call from any SIP and sends that to PSTN/mobile, so you can dial someone on mobile phone and have speech translated

Trunk has TRUNK IN and TRUNK OUT sections, so you can setup Call Flow on any of them or both. We will setup speech translation on outbound settings, in both directions

1. Like in previous steps 1-5, create a B leg Call Flow with actions to translate speech

1. Enable external calls on your Trunk. Select from your numbers, this will be a number presented as a Caller on all calls through this Trunk

1. Select your Call Flow in TRNUK OUT settings of your Trunk

✅ You now have created a Trunk that translates speech on all calls to public telephony