Initial commit

app.py

@@ -1,6 +1,8 @@
 import gradio as gr
 import json
 import torch
+import numpy as np
+
 from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
 
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
@@ -28,25 +30,11 @@ pipe = pipeline(
     device=device,
 )
 
-
-def process_audio(audio_file):
-    # In this example, let's just return a hardcoded array of JSON objects
-    output_data = [
-        {"label": "cat", "confidence": 0.8},
-        {"label": "dog", "confidence": 0.7},
-        {"label": "bird", "confidence": 0.6}
-    ]
-    return json.dumps(output_data)
-
-
 def process(audio):
-    # Read audio data from the file
-    # with open(audio.name, 'rb') as f:
-    #     audio_data = f.read()
     audio_data, audio_filename = audio
+    audio_data = np.frombuffer(audio_data.read(), dtype=np.int16)
 
-
-    # Process the audio data
+    print(audio_data)
     result = pipe(audio_data)['chunks']
     for item in result:
         item['timestamp'] = list(item['timestamp'])