smmin21/actemo-server

2024 GDSC Solution Challenge

Table of Contents

• 🔮 Features
• 🛳 Running ACTion with Docker
  • 🚀 Make Your Own API Server
    • 1 API Request Example
    • 2 API Response Example
• 🧩 Just Want to Test the Model
  • Execute Tests
• 🔗 Link to Other Parts

---

🔮 Features

ACTemo's ACTion feature enables users to choose the emotional category they wish to practice and receive real-time feedback on the accuracy of their expression in the audio input. The feedback, provided in grades like 'A', 'B', and 'C', evaluates the expression's precision, ensuring users can refine their emotional portrayal skills instantly.

🛳 Running ACTion with Docker

ACTemo provides a self-hosted version with Dockerfile. This allows you to deploy your own speech-emotion-evaluation model in just a few minutes, even without prior expertise.

🚀 Make Your Own API Server

We provide a Dockerfile for deploying the ACTemo's ACTion function on your own private device. Use the following command to start:

# 1. Build docker image
$ docker build -f SER_model/Dockerfile.prod -t actemo:prod .

# 2. Run!
$ docker run -it --rm --name actemo-prod -p 8000:8000 actemo:prod

1 API Request Example

To make a request to the API server, use the following endpoint:

• Method: POST
• Host: http://localhost:8000
• URL: /predict

In the request body, include the following keys and values:

• Attach your file to the "file" key.
• Specify one of the following emotions as the value for the "emotion" key:

'Angry', 'Distressed', 'Content', 'Happy', 'Excited', 'Joyous', 'Depressed', 'Sad', 'Bored', 'Calm', 'Relaxed', 'Sleepy', 'Afraid', 'Surprised'

Tip

We support all audio formats compatible with ffmpeg, such as m4a, mp3, caf, and more.

2 API Response Example

The response is returned in JSON format, with each audio assigned a grade, such as 'A', 'B', and 'C'.

---

🧩 Just Want to Test the Model

If you simply want to train or test the model without deploying the API server, follow these steps:

# 1. Build docker image
$ docker build -f SER_model/Dockerfile.dev -t actemo:dev .

# 2. Run!
$ docker run -it --rm --name actemo-dev -p 8000:8000 -v $(pwd)/SER_model:/root/code actemo:dev /bin/bash

Note

After running the Docker container using the provided command, you'll have access to the bash shell within the container environment.

Execute Tests

Inside the Docker container's bash shell, execute the following command to run the test:

$ python test.py

You will be prompted with the following messages:

1. "Enter the emotion: ": Input one of the listed emotions.

   'Angry', 'Distressed', 'Content', 'Happy', 'Excited', 'Joyous', 'Depressed', 'Sad', 'Bored', 'Calm', 'Relaxed', 'Sleepy', 'Afraid', 'Surprised'
   

2. "Enter the path of the audio file: ": Input the path of the audio file containing the practiced emotional expression.

Tip

This process will continue indefinitely. If you wish to stop, simply press Ctrl + C.

🔗 Link to Other Parts

If you want to explore other parts of the project, feel free to navigate to:

• Overall Project Repository: For the complete project repository, including all parts, visit the GitHub repository.
• Frontend Development: Explore the frontend development aspects of the project, including UI design and user interaction, implemented using Flutter.
• Backend Development: Dive into the backend development aspects of the project, including API implementation and database management.