Victor Shisterov

Voice API Product Manager

Victor is a product manager for Vonage Voice API with seven years of experience in the telecom industry, and a software developer since his childhood. He is passionate about making technically comp...

[Read more]
< Announcement />

Voice API Speech Recognition Now In General Availability

Last updated on Nov 20, 2020

We’re happy to announce that Speech Recognition (ASR) is now generally available! Here is the summary of improvements we have made during the Beta stage based on valuable feedback:

Call ID Is Now Optional

Unlike DTMF input, call (also known as leg) ID was a mandatory parameter for ASR. That was a bit inconvenient since you had to construct the NCCO dynamically on the fly.

Now the uuid parameter is optional, with the first leg in the call as the default, which fits the majority of ASR use-cases like IVR or voice bots. These use cases typically have a single leg in the call, either inbound from PSTN to the application, or outbound from application to the PSTN phone number.

You can still specify the leg explicitly, which could be useful in more complex scenarios.

Input Type As a Parameter

To configure the input action to accept DTMF tones only, speech only, or both, previously, you had to provide dtmf and/or speech objects respectively even if you don’t want to set any custom settings for any of them. So the default case required that you had the input action presented in the following way:

[
  {
      "action": "input",
      "dtmf": { 
      },
      "speech": {
          "uuid": "0a41d330-853b-4294-8cbb-69e8e65dc9d4"
      }
  }
]

We introduced a new parameter called type, which allows you to explicitly set what type of input action you want: [ "dtmf" ], [ "speech" ] or [ "dtmf", "speech" ] in the case of both. Keeping in mind that the uuid for speech is optional now, the NCCO object for both DTMF and ASR activated now looks as concise as:

[
  {
      "action": "input",
      "type": [ "dtmf", "speech" ]
  }
]

You can set up custom DTMF/ASR parameters as before with dtmf/speech objects respectively. For backward compatibility, the previous approach of the default DTMF input scenario is still supported.

Full SDK Support

All the available Server SDKs now support ASR.

With these improvements, converting your DTMF IVR to a natural speech voice assistant or creating one from scratch is super easy. Check out our ASR guide, NCCO reference, and Voice bot tutorial to learn more.

We never stop improving and enhancing our API and the platform, so we look forward to more feedback and your incredible apps!