# Requirements

## STT Models

### openai-whisper https://github.com/openai/whisper

#### Installation
`pip install -U openai-whisper` will install

```
mpmath, urllib3, tqdm, sympy, regex, nvidia-nvtx-cu12, nvidia-nvjitlink-cu12,
nvidia-nccl-cu12, nvidia-curand-cu12, nvidia-cufft-cu12, nvidia-cuda-runtime-cu12,
nvidia-cuda-nvrtc-cu12, nvidia-cuda-cupti-cu12, nvidia-cublas-cu12, networkx,
MarkupSafe, llvmlite, fsspec, filelock, charset-normalizer, certifi, triton,
requests, nvidia-cusparse-cu12, nvidia-cudnn-cu12, numba, jinja2, tiktoken,
nvidia-cusolver-cu12, torch, openai-whisper
```

#### Models

| Multi Langual Model | Download Size | VRAM Requirement | Relative Speed |
|---------------------|---------------| ---------------- |----------------|
| Tiny                | 70 MB         | ~1 GB | ~32x           |
| Base                | 140 MB        | ~1 GB | ~16x           | 
| Small               | 460 MB        | ~2 GB | ~6x            |
| Medium              | 1.4 GB        | ~5 GB | ~2x            |
| Large               | 2.9 GB        | ~10 GB | ~1x            |