Dynamic version of Amazon Transcribe Post Call Analytics

Question

Source: https://github.com/aws-samples/amazon-transcribe-post-call-analytics

Found the AWS solution in the source link.

By default, its behavior is configured through the inputs in the stack before you deploy it.
It takes audio files uploaded to an input bucket as input to a lambda function that starts a step-function workflow.

The confusing part is that it currently has a lot of variables in the [config file](https://github.com/aws-samples/amazon-transcribe-post-call-analytics/blob/develop/pca-server/src/pca/pcaconfiguration.py) which stores them in ssm.

Wondering how I should go about making the input and behavior of the application more dynamic? Instead of uploading a file to trigger the workflow, have a normal lambda that points to a location for a file. Or specify what transcribe options are used etc. Do any of the configs need to be dynamically accepted as inputs?

Answer

If you want to customize the solution initiation, you could write a custom Lambda function that is invoked in several ways (EventBridge, external app using SDK invoking a function URL, etc). Inside your custom Lambda, you could update the SSM Parameter Store values as you want using the SDK, and then invoke the Step Function workflow similarly.  As to the usage of the different config parameters, the linked blog below has a very detailed description of them where you can consider the impact of changing each of them.
[https://aws.amazon.com/blogs/machine-learning/post-call-analytics-for-your-contact-center-with-amazon-language-ai-services/]()

Dynamic version of Amazon Transcribe Post Call Analytics

関連するコンテンツ