Azure Text-to-Speech Service

For the Chinese version of this document, please refer to 简体中文.

This project provides a RESTful API for converting text to speech using Microsoft's Azure Cognitive Services. The service is built using Flask and can be containerized using Docker.

Features

• Convert text to speech using Microsoft Azure Cognitive Services.
• Specify voice model, speech style, and speech rate.
• Containerized using Docker for easy deployment.
• Configurable using environment variables.
• Logs are available for monitoring and debugging.

Getting Started

Prerequisites

• Python 3.10
• Docker
• Azure Cognitive Services subscription key and region

Installation

1. Clone the repository:

   git clone https://github.com/zhufengme/azure-tts-service.git
   cd azure-tts-service

2. Install dependencies (for local development):

   pip install Flask requests

Configuration

Set the following environment variables:

• AZURE_SUBSCRIPTION_KEY: Your Azure subscription key.
• AZURE_REGION: Your Azure region.

Running the Service Locally

1. Set environment variables:

   export AZURE_SUBSCRIPTION_KEY=your_azure_subscription_key
   export AZURE_REGION=your_azure_region

2. Run the Flask application:

   python app.py

Building and Running with Docker

Building the Docker Image

1. Build the Docker image:

   docker build -t azure-tts-service .

2. Run the Docker container:

   docker run -d -p 5000:5000 --name azure-tts-service -e AZURE_SUBSCRIPTION_KEY=your_azure_subscription_key -e AZURE_REGION=your_azure_region azure-tts-service

Running from DockerHub

You can pull the Docker image from DockerHub and run it directly.

1. Pull the Docker image:

   docker pull andiezhu/azure-tts-service

2. Run the Docker container:

   docker run -d -p 5000:5000 --name azure-tts-service -e AZURE_SUBSCRIPTION_KEY=your_azure_subscription_key -e AZURE_REGION=your_azure_region andiezhu/azure-tts-service

API Usage

Endpoint

• POST /tts

Request Body

• text (string, required): The text to be converted to speech.
• voice_name (string, optional): The voice model to be used. Default is en-US-AriaNeural.
• style (string, optional): The speech style to be used. Default is general.
• rate (string, optional): The speech rate. Default is 0%.

Example Request

curl -X POST http://localhost:5000/tts \
     -H "Content-Type: application/json" \
     -d '{
           "text": "Hello, this is a test of Microsoft Text to Speech service.",
           "voice_name": "en-US-AriaNeural",
           "style": "cheerful",
           "rate": "10%"
         }' --output output.wav

Health Check

The service provides a health check endpoint:

• GET /health

Example Request

curl http://localhost:5000/health

Logging

Logs are available for monitoring and debugging. By default, logs are written to both the console and a file named app.log.

Viewing Logs

To view logs, you can use the docker logs command:

docker logs -f azure-tts-service

If you configured logging to a file, you can view the logs using:

tail -f app.log

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Flask
• Requests
• Microsoft Azure Cognitive Services