Create app.py

app.py

@@ -0,0 +1,43 @@
+import streamlit as st
+import torchaudio
+import torch
+from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
+
+# Load the fine-tuned model and processor
+model_name_or_path = "./fine-tuned-uzbek-stt"  # Replace with your model's path
+processor = Wav2Vec2Processor.from_pretrained(model_name_or_path)
+model = Wav2Vec2ForCTC.from_pretrained(model_name_or_path)
+
+# Function to preprocess and transcribe audio
+def preprocess_audio(file):
+    speech_array, sampling_rate = torchaudio.load(file)
+    
+    # Resample to 16 kHz if necessary
+    if sampling_rate != 16000:
+        resampler = torchaudio.transforms.Resample(orig_freq=sampling_rate, new_freq=16000)
+        speech_array = resampler(speech_array)
+    
+    speech_array = speech_array.squeeze().numpy()
+    return speech_array
+
+def transcribe_audio(speech_array):
+    input_values = processor(speech_array, return_tensors="pt", sampling_rate=16000).input_values
+    with torch.no_grad():
+        logits = model(input_values).logits
+    predicted_ids = torch.argmax(logits, dim=-1)
+    transcription = processor.decode(predicted_ids[0])
+    return transcription.replace("[UNK]", "'")
+
+# Streamlit interface
+st.title("Speech-to-Text Transcription App")
+st.write("Upload an audio file to transcribe.")
+
+audio_file = st.file_uploader("Upload an audio file", type=["wav", "mp3"])
+
+if audio_file is not None:
+    # Preprocess and transcribe
+    speech_array = preprocess_audio(audio_file)
+    transcription = transcribe_audio(speech_array)
+    
+    st.write("Transcription:")
+    st.text(transcription)