Update app.py

app.py

@@ -1,20 +1,19 @@
-import os, time, requests, gradio as gr
-
-print("Gradio version:", gr.__version__)
+import os
+import time
+import requests
+import streamlit as st
 
 # ---------- Environment Variables ----------
 ASR_API_URL = os.getenv("ASR_API_URL")
-AUTH_TOKEN  = os.getenv("AUTH_TOKEN")
+AUTH_TOKEN = os.getenv("AUTH_TOKEN")
+
 if not ASR_API_URL or not AUTH_TOKEN:
-    print("⚠️  ASR_API_URL or AUTH_TOKEN is not set; API calls will fail.")
+    st.warning("⚠️ ASR_API_URL or AUTH_TOKEN is not set. API calls will fail.")
 
 # ---------- Core Transcription Function ----------
-def transcribe_audio(file_path: str | None):
-    if file_path is None:
-        return "Audio cleared.", "", None
-
+def transcribe_audio(file_obj):
     if not ASR_API_URL or not AUTH_TOKEN:
-        return "❌ Error: ASR_API_URL or AUTH_TOKEN is not set.", "", file_path
+        return "❌ Error: ASR_API_URL or AUTH_TOKEN is not set.", ""
 
     headers = {
         "accept": "application/json",
@@ -23,101 +22,67 @@ def transcribe_audio(file_path: str | None):
 
     start = time.time()
     try:
-        with open(file_path, "rb") as f:
-            file_name = os.path.basename(file_path)
-            # Adjusted mime type slightly - more general
-            files = {"file": (file_name, f, "audio/wav")} # Use audio/wav or audio/mpeg as appropriate for your API
-            resp = requests.post(ASR_API_URL, headers=headers, files=files, timeout=120)
-    except requests.exceptions.Timeout:
-         return f"❌ Error: Request timed out after 120 seconds.", "", file_path
+        files = {"file": ("audio.wav", file_obj, "audio/wav")}
+        resp = requests.post(ASR_API_URL, headers=headers, files=files, timeout=120)
     except Exception as e:
-        return f"❌ Error during API call or file handling: {e}", "", file_path
+        return f"❌ Error while calling ASR API: {e}", ""
 
     elapsed = time.time() - start
     if resp.status_code == 200:
-        try:
-            data = resp.json()
-            text = data.get("transcription", "No transcription returned.")
-            processing_time = data.get('time', elapsed)
-            return text, f"{processing_time:.2f} s", file_path
-        except requests.exceptions.JSONDecodeError:
-            return f"❌ Error: Could not decode JSON response. Status: {resp.status_code}, Response: {resp.text}", "", file_path
-    else:
-        return f"❌ Error: API returned status {resp.status_code}. Response: {resp.text}", "", file_path
+        data = resp.json()
+        text = data.get("transcription", "No transcription returned.")
+        return text, f"{data.get('time', elapsed):.2f} s"
+    return f"❌ Error: {resp.status_code}, {resp.text}", ""
 
+# ---------- UI ----------
+st.set_page_config(page_title="Gooya ASR v1.4", layout="centered")
 
-# ---------- Styling ----------
 VIOLET_MAIN = "#7F3FBF"
 VIOLET_LIGHT = "#C3A6FF"
+st.markdown(
+    f"""
+    <h1 style="background: linear-gradient(90deg, {VIOLET_MAIN}, {VIOLET_LIGHT}); color: white; padding: 20px; border-radius: 12px; text-align: center;">
+        Gooya ASR v1.4
+    </h1>
+    """,
+    unsafe_allow_html=True
+)
+
+tab1, tab2 = st.tabs(["🎤 Record from Microphone", "📁 Upload Audio File"])
+
+with tab1:
+    audio_file = st.audio_input("🎙️ Record audio from microphone", type="wav")
+
+with tab2:
+    uploaded_file = st.file_uploader("📂 Upload audio file (wav/mp3)", type=["wav", "mp3"])
+
+col1, col2 = st.columns(2)
+with col1:
+    btn_transcribe = st.button("Transcribe", type="primary")
+with col2:
+    btn_clear = st.button("Clear")
+
+# ---------- Main Logic ----------
+if btn_transcribe:
+    file_to_process = uploaded_file if uploaded_file else audio_file
+    if file_to_process:
+        with st.spinner("⏳ Processing..."):
+            transcription, elapsed = transcribe_audio(file_to_process)
+            st.text_area("📝 Transcription", transcription, height=150)
+            if elapsed:
+                st.info(f"⏱️ Processing Time: {elapsed}")
+    else:
+        st.warning("Please upload or record an audio file first.")
 
-custom_css = f"""
-#gooya-title {{
-  color:#fff;
-  background:linear-gradient(90deg,{VIOLET_MAIN} 0%,{VIOLET_LIGHT} 100%);
-  border-radius:12px;padding:20px 10px;margin-bottom:12px;
-}}
-.gooya-badge {{
-  display:inline-block;background:{VIOLET_MAIN};color:#fff;
-  border-radius:16px;padding:6px 16px;font-size:.97rem;margin-top:4px;
-}}
-"""
+if btn_clear:
+    st.experimental_rerun()
 
-# ---------- UI ----------
-with gr.Blocks(css=custom_css, title="Gooya ASR v1.4") as demo:
-    gr.Markdown("# Gooya ASR v1.4 Transcription", elem_id="gooya-title")
-    with gr.Row():
-        with gr.Column():
-            audio_input = gr.Audio(
-                label="Audio Input (upload or record, up to 30 s)",
-                type="filepath",
-                sources=["upload", "microphone"],
-            )
-        with gr.Column():
-            processing_time_tb = gr.Textbox(
-                label="⏱️ Processing Time",
-                interactive=False,
-                elem_classes="gooya-badge",
-            )
-            transcription_tb = gr.Textbox(
-                label="📝 Transcription",
-                lines=5,
-                show_copy_button=True,
-                placeholder="The transcription will appear here...",
-                elem_id="gooya-textbox",
-            )
-
-    with gr.Row():
-        btn_transcribe = gr.Button("Transcribe", variant="primary")
-        btn_clear      = gr.Button("Clear",      variant="secondary")
-
-    gr.Markdown(
-        """
-**Guidelines**
-- Maximum audio length: **30 seconds**
-- Audio content should be in Persian.
-- Both transcription and processing time are displayed upon completion.
-- See the [Persian ASR Leaderboard](https://huggingface.co/spaces/navidved/open_persian_asr_leaderboard) for benchmarks.
-"""
-    )
-
-    # ---------- Callbacks ----------
-    btn_transcribe.click(
-        fn=transcribe_audio,
-        inputs=[audio_input],
-        outputs=[transcription_tb, processing_time_tb, audio_input],
-    )
-
-    def clear_all():
-        return "", "", None
-
-    btn_clear.click(
-        fn=clear_all,
-        inputs=None,
-        outputs=[transcription_tb, processing_time_tb, audio_input],
-    )
-
-# ---------- Launch ----------
-if __name__ == "__main__":
-    # On Hugging Face Spaces, share=True is not needed and causes a warning.
-    # debug=True might also cause issues in some restricted environments, remove if necessary.
-    demo.queue().launch(debug=False) # <-- Reverted share=True, optionally removed debug=True
+st.markdown("""
+---
+### Guidelines
+- Maximum audio length: 30 seconds  
+- Audio content should be in Persian  
+- Both transcription and processing time will be displayed  
+
+🔗 [View the Persian ASR Leaderboard](https://huggingface.co/spaces/navidved/open_persian_asr_leaderboard)
+""")