Audio Transcription

Audio transcription tool using Nvidia's latest model Parakeet RNNT 1.1B (en). This model is jointly developed by NVIDIA NeMo and Suno.ai teams. It is an XXL version of FastConformer Transducer [1] (around 1.1B parameters) model. See the model architecture section and NeMo documentation for complete architecture details.

• Model
• Nvidia NeMo

Python package installation steps:

Note: if you are not using uv for the python environment management then just remove the uv word from the following package installation steps:

uv pip install gradio
uv pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu128
uv pip install "nemo_toolkit[all]"

Running the application

gradio main.py