metaidigitstt 0.1.2

Speech Recognition Library

MetaidigitSTT

MetaidigitSTT is a lightweight Speech-to-Text (STT) recognition module utilizing Groq's Whisper model for high-accuracy transcription. It enables real-time audio recording and transcription using Python.

Features

• Records audio via microphone and saves it as an MP3 file.
• Uses Groq's Whisper-Large-V3 model for transcription.
• Simple and easy-to-use API.

Installation

Ensure you have Python 3.10 or higher installed.

Using Pip

pip install -r requirements.txt

Required Dependencies

• ffmpeg
• SpeechRecognition
• pyaudio
• pydub
• groq
• python-dotenv

You may need to install ffmpeg manually:

# Ubuntu/Linux
sudo apt install ffmpeg

# macOS
brew install ffmpeg

# Windows
choco install ffmpeg

Usage

1. Recording Audio

The record_audio function records audio from the microphone and saves it as an MP3 file.

from metaidigitstt import record_audio

record_audio("audio/test.mp3", timeout=20, phrase_time_limit=10)

2. Transcribing Audio

The transcribe_with_groq function transcribes recorded audio using Groq's Whisper model.

from metaidigitstt import transcribe_with_groq

GROQ_API_KEY = "your_api_key_here"
stt_model = "whisper-large-v3"
audio_filepath = "audio/test.mp3"

transcription = transcribe_with_groq(stt_model, audio_filepath, GROQ_API_KEY)
print(transcription)

Environment Variables

Set up your API key in a .env file:

GROQ_API_KEY=your_api_key_here

Setup for Development

Install Locally

pip install .

Running Tests

Ensure dependencies are installed and run:

pytest tests/

Packaging & Deployment

To package the module:

python setup.py sdist bdist_wheel

To install the package locally:

pip install .

License

This project is licensed under the MIT License.

Author

Suhal Samad