Releases: SERaaS/SERaaS-API-Service
Iteration 4 Release
Iteration Count: 4
Iteration Deadline: 10th of March, 2020
This iteration deliverable provides the implementation of the following issues;
- #11 Integrate authentication functionalities to API endpoints from User Management Service
- #10 Store API Query data to User Management Service upon usage of API endpoints
This iteration of development aims to finalize the main development of this service, by integrating the authentication functionalities developed in the User Management Service to this API. Now by default, the usage of an API endpoint will require an User ID to allow for execution, and then the resulting input and output metadata will be stored in the User Management Service automatically for visualisation in the Web Platform later.
The only changes in this service from now on will just be refactoring the code and potentially integration with AWS Lambda to allow for faster SER computation.
Iteration 3 Release
Iteration Count: 3
Iteration Deadline: 25th of February, 2020
This iteration deliverable provides the implementation of the following issues;
- #7 Allow for easier Python deployment via requirements.txt file
- #8 Support showing only specific emotions in API endpoint
- #9 Support periodic query of emotions in API endpoint
This iteration of development aims to add more configuration parameters a user can use as part of the API endpoint for the service. Such parameters include the ability to limit the available emotions outputted, and also the ability to process the input audio file in chunks. In addition, a requirements.txt file was also created to simplify the deployment of the service later on in Amazon Web Services.
The following iterations will aim to improve the performance of the SER model itself.
Iteration 2 Release
Iteration Count: 2
Iteration Deadline: 11th of February, 2020
This iteration deliverable provides the implementation of the following issues;
- #4 API endpoint to classify an emotion
- #5 API endpoint to output emotional statistics
- #6 Classifier should now return probabilities rather than result
This iteration of development provides the core deliverable for this API; an API endpoint to output the emotional statistics from an audio file. This was previously followed by an implementation of an API endpoint to classify an emotion, this has now since been removed after its concept was implemented and proven a success. By changing the demo Python file to perform the SER, the emotional statistics output was achieved.
The following iterations will aim to improve the API endpoint and the performance of the SER model itself.
Iteration 1 Release
Iteration Count: 1
Iteration Deadline: 28th of January, 2020
This iteration deliverable provides the implementation of the following issues;
These deliverables setup the baseline for the development of the API for outputting emotional statistics from an audio file input. The next iterations will implement the core API functionality, which involves modifying the demo classifer to output probabilities of results rather than a final classification (#6).
Note that the following issue has been moved to the next iteration, which is due on the 11th of February, 2020.
- #4 API endpoint to classify an emotion