Create app.py

app.py

@@ -0,0 +1,35 @@
+import gradio as gr
+import torch
+import librosa
+from transformers import WhisperProcessor, WhisperForConditionalGeneration
+
+# Load the Whisper model
+model_id = "ivrit-ai/whisper-large-v3-turbo"
+processor = WhisperProcessor.from_pretrained(model_id)
+model = WhisperForConditionalGeneration.from_pretrained(model_id)
+
+# Move model to GPU if available
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model.to(device)
+
+# Function to transcribe Hebrew audio
+def transcribe(audio):
+    waveform, sr = librosa.load(audio, sr=16000)  # Convert to 16kHz
+    input_features = processor(waveform, sampling_rate=16000, return_tensors="pt").input_features.to(device)
+
+    with torch.no_grad():
+        predicted_ids = model.generate(input_features)
+
+    transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
+    return transcription
+
+# Create the Gradio Interface
+iface = gr.Interface(
+    fn=transcribe,
+    inputs=gr.Audio(source="upload", type="filepath"),
+    outputs="text",
+    title="Hebrew Speech-to-Text (Whisper)",
+    description="Upload a Hebrew audio file and receive a transcription.",
+)
+
+iface.launch()