If you see Unknown Speaker(s) either on your transcript, or in the Participants section of the Recording Details page, this means that Conversations was unable to detect the speaker(s).

But why? To answer this, let’s walk through the process Conversations uses to detect speakers.

When a recording is processed, Conversations analyzes the transcription to differentiate between speakers, identify talk-times…etc. There are two different types of transcripts: 

• Native transcripts from the call provider, i.e Zoom, Webex, Microsoft Teams
• AWS Transcripts, i.e. transcripts from a manual upload or a bot capture

Native transcripts contain data that Conversations uses to match speakers with Salesloft users. Transcripts from the native call provider include full speaker detection and should include the speaker(s) name(s).

Transcripts that come from a manual upload or a bot captured meeting are processed via our transcription provider, Amazon Transcribe (AWS). AWS transcripts include partial speaker detection that differentiates between speakers, but may not identify speaker names. 

To ensure that your transcripts include speaker names, we recommend the following:

• Use an API Integrations to capture meetings in Conversations. Our API Integrations use the video conferencing platform’s recording capabilities, thus, providing Conversations with a native transcript that can be matched to Salesloft users.
  • When using an API Integration, be sure to enable any audio transcript settings from your account in the video conferencing platform. This will ensure that a transcript is created.