Update app.py

app.py

@@ -3,52 +3,40 @@ import torch
 import torchaudio
 from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
 
-# Load the base model and processor outside the function to avoid reloading
+# Load the model
 model_id = "facebook/wav2vec2-base-960h"
 processor = Wav2Vec2Processor.from_pretrained(model_id)
 model = Wav2Vec2ForCTC.from_pretrained(model_id)
 
-# Transcription function with optimization
 def transcribe(audio_file, progress=gr.Progress()):
     if audio_file is None:
         return "⚠️ No audio received."
 
-    print(f"📥 Received file path: {audio_file}")  # ✅ Log for debugging
-
-    progress(0, desc="Loading audio...")
     waveform, sample_rate = torchaudio.load(audio_file)
 
     if sample_rate != 16000:
-        progress(0.3, desc="Resampling audio...")
         waveform = torchaudio.functional.resample(waveform, orig_freq=sample_rate, new_freq=16000)
         sample_rate = 16000
 
     if waveform.shape[0] > 1:
-        progress(0.5, desc="Converting to mono...")
         waveform = waveform.mean(dim=0).unsqueeze(0)
 
-    progress(0.7, desc="Processing audio...")
     input_values = processor(waveform.squeeze().numpy(), sampling_rate=sample_rate, return_tensors="pt").input_values
 
     with torch.no_grad():
-        progress(0.8, desc="Transcribing...")
         logits = model(input_values).logits
 
     predicted_ids = torch.argmax(logits, dim=-1)
     transcription = processor.batch_decode(predicted_ids)[0]
-
-    progress(1.0, desc="Done!")
     return transcription.lower()
 
-# Gradio UI with POST API support
 demo = gr.Interface(
     fn=transcribe,
     inputs=gr.Audio(sources=["microphone"], type="filepath", label="🎤 Speak now"),
     outputs=gr.Textbox(label="📝 Transcription"),
     title="Wav2Vec2 Speech Transcription",
-    description="Speak into the microphone and get a transcription using Wav2Vec2-base (via Hugging Face Transformers).",
-    allow_flagging="never"
+    description="Speak into the microphone and get a transcription using Wav2Vec2-base.",
+    flagging_mode="never"
 )
 
-# ✅ Enable API mode to allow POST requests
-demo.launch(api=True)
+demo.launch()