Batch recording API and metadata request for speech analytics

[kylecarter]

Hey all! First time posting here.

We are working on an application to pull batch recordings and metadata for analysis. Could you provide some guidance on the appropriate API's to use to pull conversation recordings in batch along with their associated metadata (conversation detail such as agent, dispositioning, queue, etc)?

Our goal is to query for previous day 's interactions and automate into our SA application.

Also, are recordings returned as one file or multiple per conversation (transfers, etc.)? We'd prefer single files if configurable that way.

Appreciate any help the community can provide.

Kyle

[tim.smith]

You can export recordings in bulk using the AWS S3 recording bulk actions integration. If you're only doing selective exporting, you can use POST /api/v2/recording/batchrequests to provide a specific list of recordings and get temporary download URLs for them.

@kylecarter

Also, are recordings returned as one file or multiple per conversation (transfers, etc.)? We'd prefer single files if configurable that way.

There is always the possibility to have multiple recordings per conversation for a variety of reasons. One you can control with a setting is dual channel recording on the trunk.

[kylecarter]

Tim - thanks for the reply. Tell me if I'm thinking about this correctly process wise:

1. Query for a range of conversation IDs (ex. occuring on 12/1/2021)
2. Batch recordings for conversation IDs returned above https://developer.genesys.cloud/api/rest/v2/recording/#post-api-v2-recording-batchrequests
3. Pull metadata for conversation IDs returned in #1 with https://developer.genesys.cloud/api/rest/v2/recording/#get-api-v2-conversations--conversationId--recordingmetadata

I think building it this way would give us the flexibility to change cadence or expand the scope of the pull as needed. The only question I'd have remaining would be how do we get conversation IDs for a date range (#1 above)?

Let me know if I'm off base here.

KC

[tim.smith]

You'd use the analytics conversation detail query for #1. However, I'd strongly recommend using the AWS integration as it accepts the analytics query and does all of those steps for you and has a much greater throughput potential due to rate limiting you're going to hit with the batchrequests endpoints. The batch requests are meant for targeted downloading of individual recordings and have rate limits applied as such. The AWS integration is specifically made for high-volume exports and then you can hammer away downloading from S3 at any rate AWS can handle.