Shridhar7-8/swahili_asr_translation
Domain:
natural language processingRecord type:
software
Creator:
Shr
Host:
# Swahili ASR and Translation API
This FastAPI application provides an endpoint for transcribing Swahili speech from .wav files and translating the transcription into English.
## Features
- **Automatic Speech Recognition (ASR)** using speechbrain's wav2vec2 model.
- **Translation** from Swahili to English using Hugging Face Transformers.
- **FastAPI-based web service** with a `/docs` interface for testing.
## Installation
To get started, clone the repository and set up your environment.
### 1. **Clone the repository**:
```bash
git clone github.com
cd swahili_asr_translation
```
### 2. **Set up a virtual environment**
```bash
python -m venv venv
```
### 3. Activate the virtual environment
- On macOS/Linux:
```bash
source venv/bin/activate
```
- On Windows:
```bash
venv\Scripts\activate
```
### 4. Install the required dependencies
```bash
pip install -r requirements.txt
```
## Usage
### Running the API server:
Run the following command to start the FastAPI application:
```bash
uvicorn swahili_asr_translation:app --host 0.0.0.0 --port 8021 --reload
```
Once the server is running, you can access the API documentation at:
```bash
localhost
```
## API Endpoints
### `POST /transcribe-and-translate/`
Uploads a `.wav` file, transcribes it to Swahili text, and translates it into English.
#### Request:
- **File**: A `.wav` audio file.
#### Response (JSON):
```json
{
"audio_file": "example.wav",
"transcription": "Hii ni sauti ya majaribio",
"translation": "This is a test audio"
}
```
## Deployment
For production environments, it is recommended to run the application with multiple workers for better scalability.
If you're running the app from a script, you can use the following setup in your `swahili_api.py`:
```python
if __name__ == "__main__":
import uvicorn
uvicorn.run("swahili_asr_translation:app", host="0.0.0.0", port=8021, workers=4)
```