Create app.py

app.py

@@ -0,0 +1,56 @@
+from nemo.collections.asr.models import NeuralDiarizer
+from nemo.collections.asr.parts.utils.speaker_utils import rttm_to_labels
+import gradio as gr
+import torch
+import pandas as pd
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+model = NeuralDiarizer.from_pretrained("diar_msdd_telephonic").to(device)
+
+
+def run_diarization(path1):
+    annotation = model(path1)
+    rttm=annotation.to_rttm()
+    df = pd.DataFrame(columns=['start_time', 'end_time', 'speaker'])
+    for idx,line in enumerate(rttm.splitlines()):
+        split = line.split()
+        start_time, end_time, speaker = split[3], split[4], split[7]
+        df.loc[idx] = start_time, end_time, speaker
+    return df 
+	
+inputs = [
+    gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Input Audio"),
+]
+output = gr.outputs.Dataframe()
+
+
+description = (
+    "This demonstration will perform offline speaker diarization on an audio file using nemo"
+)
+
+article = (
+    "<p style='text-align: center'>"
+    "<a href='https://huggingface.co/nvidia/speakerverification_en_titanet_large' target='_blank'>🎙️ Learn more about TitaNet model</a> | "
+    "<a href='https://arxiv.org/pdf/2110.04410.pdf' target='_blank'>📚 TitaNet paper</a> | "
+    "<a href='https://github.com/NVIDIA/NeMo' target='_blank'>🧑‍💻 Repository</a>"
+    "</p>"
+)
+examples = [
+    "data/id10270_5r0dWxy17C8-00001.wav",
+]
+
+interface = gr.Interface(
+    fn=run_diarization,
+    inputs=inputs,
+    outputs=output,
+    title="Offline Speaker Diarization with NeMo",
+    description=description,
+    article=article,
+    layout="horizontal",
+    theme="huggingface",
+    allow_flagging=False,
+    live=False,
+    examples=examples,
+)
+interface.launch(enable_queue=True)