Update handler.py

handler.py

@@ -1,15 +1,16 @@
 from typing import Dict
 from pyannote.audio import Pipeline
-import torch 
-import base64
-import numpy as np
+import torch
+import io
+from pyannote.audio import Audio
 
 SAMPLE_RATE = 16000
 
 class EndpointHandler():
     def __init__(self, path=""):
-        # load the model
-        self.pipeline = Pipeline.from_pretrained("KIFF/pyannote-speaker-diarization-endpoint")
+        # Load the pipeline from the model repository (using config.yaml)
+        self.pipeline = Pipeline.from_pretrained(path)
+        self.audio = Audio(sample_rate=SAMPLE_RATE, mono="downmix")
 
     def __call__(self, data: Dict[str, bytes]) -> Dict[str, str]:
         """
@@ -21,16 +22,14 @@ class EndpointHandler():
         """
         # process input
         inputs = data.pop("inputs", data)
-        parameters = data.pop("parameters", None) #  min_speakers=2, max_speakers=5
+        parameters = data.pop("parameters", None)
 
-        # decode the base64 audio data
-        audio_data = base64.b64decode(inputs)
-        audio_nparray = np.frombuffer(audio_data, dtype=np.int16)
+        # Load the audio using pyannote.audio
+        waveform, sample_rate = self.audio(io.BytesIO(inputs))
+
+        # prepare pyannote input
+        pyannote_input = {"waveform": waveform, "sample_rate": sample_rate}
 
-        # prepare pynannote input
-        audio_tensor= torch.from_numpy(audio_nparray).float().unsqueeze(0)
-        pyannote_input = {"waveform": audio_tensor, "sample_rate": SAMPLE_RATE}
-        
         # apply pretrained pipeline
         # pass inputs with all kwargs in data
         if parameters is not None:
@@ -43,5 +42,5 @@ class EndpointHandler():
             {"label": str(label), "start": str(segment.start), "stop": str(segment.end)}
             for segment, _, label in diarization.itertracks(yield_label=True)
         ]
-        
-        return {"diarization": processed_diarization}
+
+        return {"diarization": processed_diarization}