Update app.py

app.py

@@ -1,7 +1,19 @@
+import os
 import re
+import requests
 import gradio as gr
 
-# regex patterns
+# =========================
+# ElevenLabs Config
+# =========================
+
+ELEVENLABS_API_KEY = os.getenv("sk_43cdb8a03472dc061bc3468dc05777022870f95dc2c907e7")
+STT_URL = "https://api.elevenlabs.io/v1/speech-to-text"
+
+# =========================
+# Regex Cleaning
+# =========================
+
 REPEAT_WORD = re.compile(r'\b(\w+)(?:\s+\1\b)+', re.IGNORECASE)
 CHAR_STRETCH = re.compile(r'(.)\1{2,}')
 REPEAT_SYLLABLE = re.compile(r'\b(\w{1,3})(?:\s+\1\b)+', re.IGNORECASE)
@@ -12,6 +24,7 @@ def is_filler(word):
     if len(set(w)) == 1 and len(w) <= 4:
         return True
 
+    # مقطع قصير جداً
     if len(w) <= 2:
         return True
 
@@ -20,13 +33,8 @@ def is_filler(word):
 
 def clean_transcript(text):
 
-    # collapse stretched characters
     text = CHAR_STRETCH.sub(r'\1', text)
-
-    # remove repeated words
     text = REPEAT_WORD.sub(r'\1', text)
-
-    # remove repeated short syllables
     text = REPEAT_SYLLABLE.sub(r'\1', text)
 
     words = text.split()
@@ -39,22 +47,78 @@ def clean_transcript(text):
     return " ".join(filtered)
 
 
-def process(text):
-    return clean_transcript(text)
+# =========================
+# Speech To Text
+# =========================
+
+def transcribe_audio(audio_file):
+
+    if audio_file is None:
+        return "No audio uploaded", ""
+
+    headers = {"xi-api-key": ELEVENLABS_API_KEY}
+
+    with open(audio_file, "rb") as f:
+        files = {"file": f}
+        data = {
+            "model_id": "scribe_v2",
+            "enable_logging": "false"
+        }
+
+        response = requests.post(
+            STT_URL,
+            headers=headers,
+            files=files,
+            data=data
+        )
+
+    if response.status_code != 200:
+        return f"Error: {response.text}", ""
+
+    result = response.json()
+
+    text = ""
+
+    if "segments" in result:
+        for seg in result["segments"]:
+            text += seg.get("text", "") + " "
+    else:
+        text = result.get("text", "")
+
+    cleaned = clean_transcript(text)
+
+    return text, cleaned
+
+
+# =========================
+# Gradio UI
+# =========================
+
+with gr.Blocks() as demo:
+
+    gr.Markdown("# Speech To Text Cleaner")
+    gr.Markdown("Upload audio → convert to text → remove fillers")
+
+    audio_input = gr.Audio(type="filepath", label="Upload Audio")
+
+    raw_text = gr.Textbox(
+        label="Original Transcript",
+        lines=8
+    )
+
+    cleaned_text = gr.Textbox(
+        label="Cleaned Transcript",
+        lines=8
+    )
+
+    btn = gr.Button("Transcribe")
 
+    btn.click(
+        fn=transcribe_audio,
+        inputs=audio_input,
+        outputs=[raw_text, cleaned_text]
+    )
 
-demo = gr.Interface(
-    fn=process,
-    inputs=gr.Textbox(
-        lines=8,
-        placeholder="Paste transcript here..."
-    ),
-    outputs=gr.Textbox(
-        lines=8,
-        label="Cleaned transcript"
-    ),
-    title="Transcript Filler Cleaner",
-    description="Remove repeated words and speech fillers automatically"
-)
 
-demo.launch()
+if __name__ == "__main__":
+    demo.launch()