Workforce Engagement Management

Hi! 

In our company we have recently started working with Genesys Native Voice Transcripts. We have our own company data lake and export them there too for further analysis and summarisation using our own context trained LLMs. When we receive transcript files for calls with more than one agent participating we cannot understand how to distinguish between them. 

The purpose for all internal paticipants is the same "agent" and in the transcript it appears like that:

"transcripts": [

        {

            "text": "bla ... bla",

            "participantPurpose": "internal"

        },

        {

            "text": "bla ... bla",

            "participantPurpose": "external"

        }

    ],

It doesn't contain any user ids, participant ids, only a purpose which doesn't match anything in the participant list and has no unique id. 

The question is: how can we distinguish different internal/agent participants in the same transcript?

Thank you!

#VoiceTranscription #SpeechandTextAnalytics

Hi Kate,

I'm trying to source an answer for you.

Hi! 

In our company we have recently started working with Genesys Native Voice Transcripts. We have our own company data lake and export them there too for further analysis and summarisation using our own context trained LLMs. When we receive transcript files for calls with more than one agent participating we cannot understand how to distinguish between them. 

The purpose for all internal paticipants is the same "agent" and in the transcript it appears like that:

"transcripts": [

        {

            "text": "bla ... bla",

            "participantPurpose": "internal"

        },

        {

            "text": "bla ... bla",

            "participantPurpose": "external"

        }

    ],

It doesn't contain any user ids, participant ids, only a purpose which doesn't match anything in the participant list and has no unique id. 

The question is: how can we distinguish different internal/agent participants in the same transcript?

Thank you!

#VoiceTranscription #SpeechandTextAnalytics

Hi Kate,

I'm trying to source an answer for you.

Hi! 

In our company we have recently started working with Genesys Native Voice Transcripts. We have our own company data lake and export them there too for further analysis and summarisation using our own context trained LLMs. When we receive transcript files for calls with more than one agent participating we cannot understand how to distinguish between them. 

The purpose for all internal paticipants is the same "agent" and in the transcript it appears like that:

"transcripts": [

        {

            "text": "bla ... bla",

            "participantPurpose": "internal"

        },

        {

            "text": "bla ... bla",

            "participantPurpose": "external"

        }

    ],

It doesn't contain any user ids, participant ids, only a purpose which doesn't match anything in the participant list and has no unique id. 

The question is: how can we distinguish different internal/agent participants in the same transcript?

Thank you!

#VoiceTranscription #SpeechandTextAnalytics

Hi Kate,

Here is a response from the Product Team - hope it helps answer your question:

In the case of multiple agents participating in a call (e.g., transfer, etc) they can follow the same logic as the UI, which is , find the interceptions of participant information (coming from conversation details) and the transcript phrase purpose (internal and external) to identify the participants.

In the case when multiple agents are participating in the interaction at the same time (e.g., conference, etc) it is not straightforward to map to an individual agent as there are overlapping start_time_ms. In the UI we define those instances as "conference", and not individual agents.

Hi Kate,

I'm trying to source an answer for you.

Hi! 

In our company we have recently started working with Genesys Native Voice Transcripts. We have our own company data lake and export them there too for further analysis and summarisation using our own context trained LLMs. When we receive transcript files for calls with more than one agent participating we cannot understand how to distinguish between them. 

The purpose for all internal paticipants is the same "agent" and in the transcript it appears like that:

"transcripts": [

        {

            "text": "bla ... bla",

            "participantPurpose": "internal"

        },

        {

            "text": "bla ... bla",

            "participantPurpose": "external"

        }

    ],

It doesn't contain any user ids, participant ids, only a purpose which doesn't match anything in the participant list and has no unique id. 

The question is: how can we distinguish different internal/agent participants in the same transcript?

Thank you!

#VoiceTranscription #SpeechandTextAnalytics