This will build the dockerfile and create a github package that can be downloaded any time there's a push to 'main'.
Once this is in place and built, users can simply:
```
docker pull ghcr.i...
**I get an error when i run inference-cli.py with my code:**
python inference-cli.py \
--model "F5-TTS" \
--ref_audio "/home/pdnguyen/TTS/F5-TTS/notebooks/data/fine-tune/wavs/chunk_2.wav" \
--r...
This happens when input to torch.stft is too short... it prevents you from computing stft over mostly zero-padding, which would be useless.
+1 to just filtering out samples with duration less th...