FAQ

Table of Contents

General Questions

1. How do Team Voice, the Voice SDK and Scribe all work together?

Team Voice is a full-fledged application implementing our Voice and Scribe APIs behind the scenes. Companies that want instant, collaborative voice communications with automatic data extraction can use Team Voice out of the box. For firms that wish to incorporate push-to-talk or transcription functionality directly in their application, the Voice SDK and Scribe API offer the development endpoints to seamlessly integrate GreenKey’s technology into any custom workflow.


Voice

1. How do I recover my password?

In the event that you forget your login credentials, you can either select the password retrieval link located on your login screen and an email with instructions will be sent to the email on file. You can also contact your administrator to have them reset it for you. Additionally, you can contact GreenKey Support and we will reset it for you as well.

2. How do I change my password in the application?

To change your password, you will need to access Settings via the icon. - On the General page, scroll down to the bottom and select Change Password - Enter in twice to confirm and Save.

3. How do I change my audio settings?

To change your audio settings you can: - Access Settings via the icon - Access the Audio tab (middle tab between General and Preferences) - Select the desired audio device from the drop down, then Save

4. How do I test my audio settings?

To test your audio input and output device you will need to access Settings (via the icon) - Access the Audio tab (middle tab between General and Preferences) - Verify that the device you want as your input/output is correct. - For input, you should see the equalizer bar move when speaking. - For output, select the play sound noise and verify that you can hear the audible tone coming from the speaker you desire.

5. How do I create a channel?

You can create new channels and edit channels (if you are the creator), add members, and apply a business line to the channel. Having a phone number associated with the channel allows all the members in the channel to see and join any active call. Select Add channel Icon located in the top of your list next to the search field - Name Channel - Description (optional) - Add phone number (optional)- This will be the number that is associated with the channel and appear when you dial out. - Invite members- Choose team members you want to be included in this channel. - Create (Save) - Channel Name will appear on list


Scribe

1. How accurate will Scribe be on my particular accent / conversation / audio type?

Scribe is trained on hundreds of accents and thousands of hours of data, including general English conversations, financial conversations, quotes and trades, and more. On average, Scribe can achieve 85% transcription accuracy or greater on high quality audio, and 95% accuracy or greater on extracted quotes and trades. If your audio is poor in quality, out of the box accuracy will be in the 50% to 60% range. Customizing Scribe can increase accuracy on poor quality audio to around 70%. Once audio quality is improved, higher accuracy can be achieved.

2. How scalable is Scribe?

Scribe can be horizontally scaled by launching dozens of individual Scribe Docker containers in parallel to handle large transcription needs. GreenKey recommends using Kubernetes to deploy and scale Scribe, and we provide example scripts to get you up and running. Typically, real-time transcription requires 1 CPU and 1 GB RAM per active user, while delayed transcription is 3x – 4x more resource efficient.

3. Can Scribe recognize speakers within a conversation?

Scribe implements a process called “Speaker Diarization”, where the conversation is broken up into individual chunks of audio that only contain a single speaker speaking. The chunks are then “clustered” together. In a two person conversation, this means your transcript will say “Speaker 1: hi”, “Speaker 2: hello”, “Speaker 1: how are you”, and so on.

If you have a voice sample of a person you know for certain was on the call, you can feed to it Scribe and it will tell you which speaker is most likely the one you targeted. However, Scribe does not search a database of voice samples to identify all speakers on a call.

4. How well does Scribe perform on general English (non-financial) conversations versus cloud providers?

Scribe achieves around 93% accuracy on high quality general English conversations. Cloud providers typically achieve 90% - 95% accuracy on the same type of audio.

5. Does Scribe support multiple languages?

At the moment, Scribe only supports English, but other languages are in development. Contact Us for more information about a specific language release date.


Voice SDK

1. Are there any examples of how we can use the Voice SDK?

Participants are strongly advised to run the example applications and tweak or cut/paste code from that base. The example repositories should be considered reference implementations that will help participant developers know how to use the SDK.

2. What other functionality is coming to the Voice SDK?

Currently, the Voice SDK only includes access to the intercom API. In the near future, we will be opening access to the Call API (advanced calling, transfer, merge, barge), User Management API (authentication, add/edit users), and others.

3. When can we expect additional functionality in the API?

We are currently adding in functionality for both advanced calling and user management to our SDK. We expect to release user management in Q1 2018 and advanced calling in Q2 2018.

4. Can I get Transcription added to the Voice SDK?

Yes, although the Scribe API is independent of the Voice SDK, it is included in our Voice & Scribe SDK offering. Please contact a GreenKey representative to learn more.

5. What the difference between GreenKey’s API vs. SDK offering?

GreenKey’s SDK or “Software Development Kit” includes all of the components (API’s) that make up our Voice platform. Consider this, if an API is a set of building blocks that allows for the creation of something, an SDK is a full-fledged laboratory, facilitating creation far outside the scopes of what an API would allow. The API’s are actually the individual component of the functionality included in our GreenKey Voice Application. Currently these include the Intercom API, Call API, and User Management API.


In case you haven’t found the answer for your question please feel free to Contact Us, our customer support will be happy to help you.