Faster Whisper Transcription Service

Overview

This project uses the faster_whisper Python package to provide an API endpoint for audio transcription. It utilizes OpenAI's Whisper model (large-v3) for accurate and efficient speech-to-text conversion. The service is designed to be deployed on Hugging Face endpoints.

Features

Efficient Transcription: Utilizes the large-v3 Whisper model for high-quality transcription.
Multilingual Support: Supports transcription in various languages, with default language set to German (de).
Segmented Output: Returns transcribed text with segment IDs and timestamps for each transcribed segment.

Usage

import requests
import os

# Sample data dict with the link to the video file and the desired language for transcription
DATA = {
    "inputs": "<base64_encoded_audio>",
    "language": "de",
    "task": "transcribe"
}

HF_ACCESS_TOKEN = os.environ.get("HF_TRANSCRIPTION_ACCESS_TOKEN")
API_URL = os.environ.get("HF_TRANSCRIPTION_ENDPOINT")

HEADERS = {
    "Authorization": HF_ACCESS_TOKEN,
    "Content-Type": "application/json"
}

response = requests.post(API_URL, headers=HEADERS, json=DATA)
print(response)

Logging

Logging is set up to debug level, providing detailed information during the transcription process, including the length of decoded bytes, the progress of segments being transcribed, and a confirmation once the inference is completed.

Deployment

This service is intended for deployment on Hugging Face endpoints. Ensure you follow Hugging Face's guidelines for deploying model endpoints.