Added audio separation

app.py

@@ -4,6 +4,20 @@ import json
 import numpy as np
 from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
 
+# This code was omitted for deployment reasons (model is too RAM-hungry)
+
+# from speechbrain.inference.separation import SepformerSeparation as separator
+# import torchaudio
+
+# model = separator.from_hparams(source="speechbrain/sepformer-whamr16k", savedir='pretrained_models/sepformer-whamr16k')
+
+# def separate_speech(path):
+#     est_sources = model.separate_file(path=path)
+#     output_path = "output.wav"
+#     torchaudio.save(output_path, est_sources[:, :, 0].detach().cpu(), 16000)
+#     return output_path
+
+
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
 torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32