Update app.py

app.py

@@ -2,13 +2,11 @@ import gradio as gr
 from pydub import AudioSegment
 import json
 import uuid
+import aiohttp
 import asyncio
-import aiofiles
 import os
 import time
-import mimetypes
 from typing import List, Dict
-import aiohttp # برای درخواست های HTTP به Talkbot.ir
 
 # Constants
 MAX_FILE_SIZE_MB = 20
@@ -16,11 +14,11 @@ MAX_FILE_SIZE_BYTES = MAX_FILE_SIZE_MB * 1024 * 1024  # Convert MB to bytes
 
 class PodcastGenerator:
     def __init__(self):
-        self.talkbot_tts_url = "https://talkbot.ir/TTS-tkun"
-        # OpenRouter API URL از Talkbot.ir
-        self.talkbot_openrouter_api_url = "https://talkbot.ir/api/v1/chat/completions"
+        self.api_key = "sk-4fb613f56acfccf731e801b904cd89f5"
+        self.api_url = "https://talkbot.ir/api/v1/chat/completions"
+        self.tts_url = "https://talkbot.ir/TTS-tkun"
 
-    async def generate_script(self, prompt: str, language: str, api_key: str, file_obj=None, progress=None) -> Dict:
+    async def generate_script(self, prompt: str, language: str, file_obj=None, progress=None) -> Dict:
         example = """
 {
     "topic": "AGI",
@@ -32,182 +30,6 @@ class PodcastGenerator:
         {
             "speaker": 1,
             "line": "Yeah, it's definitely having a moment, isn't it?"
-        },
-        {
-            "speaker": 2,
-            "line": "It is and for good reason, right? I mean, you've been digging into this stuff, listening to the podcasts and everything. What really stood out to you? What got you hooked?"
-        },
-        {
-            "speaker": 1,
-            "line": "Honestly, it's the sheer scale of what AGI could do. We're talking about potentially reshaping well everything."
-        },
-        {
-            "speaker": 2,
-            "line": "No kidding, but let's be real. Sometimes it feels like every other headline is either hyping AGI up as this technological utopia or painting it as our inevitable robot overlords."
-        },
-        {
-            "speaker": 1,
-            "line": "It's easy to get lost in the noise, for sure."
-        },
-        {
-            "speaker": 2,
-            "line": "Exactly. So how about we try to cut through some of that, shall we?"
-        },
-        {
-            "speaker": 1,
-            "line": "Sounds like a plan."
-        },
-        {
-            "speaker": 2,
-            "line": "Okay, so first things first, AGI, what is it really? And I don't just mean some dictionary definition, we're talking about something way bigger than just a super smart computer, right?"
-        },
-        {
-            "speaker": 1,
-            "line": "Right, it's not just about more processing power or better algorithms, it's about a fundamental shift in how we think about intelligence itself."
-        },
-        {
-            "speaker": 2,
-            "line": "So like, instead of programming a machine for a specific task, we're talking about creating something that can learn and adapt like we do."
-        },
-        {
-            "speaker": 1,
-            "line": "Exactly, think of it this way: Right now, we've got AI that can beat a grandmaster at chess but ask that same AI to, say, write a poem or compose a symphony. No chance."
-        },
-        {
-            "speaker": 2,
-            "line": "Okay, I see. So, AGI is about bridging that gap, creating something that can move between those different realms of knowledge seamlessly."
-        },
-        {
-            "speaker": 1,
-            "line": "Precisely. It's about replicating that uniquely human ability to learn something new and apply that knowledge in completely different contexts and that's a tall order, let me tell you."
-        },
-        {
-            "speaker": 2,
-            "line": "I bet. I mean, think about how much we still don't even understand about our own brains."
-        },
-        {
-            "speaker": 1,
-            "line": "That's exactly it. We're essentially trying to reverse-engineer something we don't fully comprehend."
-        },
-        {
-            "speaker": 2,
-            "line": "And how are researchers even approaching that? What are some of the big ideas out there?"
-        },
-        {
-            "speaker": 1,
-            "line": "Well, there are a few different schools of thought. One is this idea of neuromorphic computing where they're literally trying to build computer chips that mimic the structure and function of the human brain."
-        },
-        {
-            "speaker": 2,
-            "line": "Wow, so like actually replicating the physical architecture of the brain. That's wild."
-        },
-        {
-            "speaker": 1,
-            "line": "It's pretty mind-blowing stuff and then you've got folks working on something called whole brain emulation."
-        },
-        {
-            "speaker": 2,
-            "line": "Okay, and what's that all about?"
-        },
-        {
-            "speaker": 1,
-            "line": "The basic idea there is to create a complete digital copy of a human brain down to the last neuron and synapse and run it on a sufficiently powerful computer simulation."
-        },
-        {
-            "speaker": 2,
-            "line": "Hold on, a digital copy of an entire brain, that sounds like something straight out of science fiction."
-        },
-        {
-            "speaker": 1,
-            "line": "It does, doesn't it? But it gives you an idea of the kind of ambition we're talking about here and the truth is we're still a long way off from truly achieving AGI, no matter which approach you look at."
-        },
-        {
-            "speaker": 2,
-            "line": "That makes sense but it's still exciting to think about the possibilities, even if they're a ways off."
-        },
-        {
-            "speaker": 1,
-            "line": "Absolutely and those possibilities are what really get people fired up about AGI, right? Yeah."
-        },
-        {
-            "speaker": 2,
-            "line": "For sure. In fact, I remember you mentioning something in that podcast about AGI's potential to revolutionize scientific research. Something about supercharging breakthroughs."
-        },
-        {
-            "speaker": 1,
-            "line": "Oh, absolutely. Imagine an AI that doesn't just crunch numbers but actually understands scientific data the way a human researcher does. We're talking about potential breakthroughs in everything from medicine and healthcare to material science and climate change."
-        },
-        {
-            "speaker": 2,
-            "line": "It's like giving scientists this incredibly powerful new tool to tackle some of the biggest challenges we face."
-        },
-        {
-            "speaker": 1,
-            "line": "Exactly, it could be a total game changer."
-        },
-        {
-            "speaker": 2,
-            "line": "Okay, but let's be real, every coin has two sides. What about the potential downsides of AGI? Because it can't all be sunshine and roses, right?"
-        },
-        {
-            "speaker": 1,
-            "line": "Right, there are definitely valid concerns. Probably the biggest one is the impact on the job market. As AGI gets more sophisticated, there're a real chance it could automate a lot of jobs that are currently done by humans."
-        },
-        {
-            "speaker": 2,
-            "line": "So we're not just talking about robots taking over factories but potentially things like, what, legal work, analysis, even creative fields?"
-        },
-        {
-            "speaker": 1,
-            "line": "Potentially, yes. And that raises a whole host of questions about what happens to those workers, how we retrain them, how we ensure that the benefits of AGI are shared equitably."
-        },
-        {
-            "speaker": 2,
-            "line": "Right, because it's not just about the technology itself, but how we choose to integrate it into society."
-        },
-        {
-            "speaker": 1,
-            "line": "Absolutely. We need to be having these conversations now about ethics, about regulation, about how to make sure AGI is developed and deployed responsibly."
-        },
-        {
-            "speaker": 2,
-            "line": "So it's less about preventing some kind of sci-fi robot apocalypse and more about making sure we're steering this technology in the right direction from the get-go."
-        },
-        {
-            "speaker": 1,
-            "line": "Exactly, AGI has the potential to be incredibly beneficial, but it's not going to magically solve all our problems. It's on us to make sure we're using it for good."
-        },
-        {
-            "speaker": 2,
-            "line": "It's like you said earlier, it's about shaping the future of intelligence."
-        },
-        {
-            "speaker": 1,
-            "line": "I like that. It really is."
-        },
-        {
-            "speaker": 2,
-            "line": "And honestly, that's a responsibility that extends beyond just the researchers and the policymakers."
-        },
-        {
-            "speaker": 1,
-            "line": "100%"
-        },
-        {
-            "speaker": 2,
-            "line": "So to everyone listening out there I'll leave you with this. As AGI continues to develop, what role do you want to play in shaping its future?"
-        },
-        {
-            "speaker": 1,
-            "line": "That's a question worth pondering."
-        },
-        {
-            "speaker": 2,
-            "line": "It certainly is and on that note, we'll wrap up this deep dive. Thanks for listening, everyone."
-        },
-        {
-            "speaker": 1,
-            "line": "Peace."
         }
     ]
 }
@@ -230,157 +52,134 @@ You are a professional podcast generator. Your task is to generate a professiona
 Follow this example structure:
 {example}
 """
-        user_prompt_text = ""
+        
+        user_prompt = ""
         if prompt and file_obj:
-            user_prompt_text = f"Please generate a podcast script based on the uploaded file following user input:\n{prompt}"
+            user_prompt = f"Please generate a podcast script based on the uploaded file following user input:\n{prompt}"
         elif prompt:
-            user_prompt_text = f"Please generate a podcast script based on the following user input:\n{prompt}"
+            user_prompt = f"Please generate a podcast script based on the following user input:\n{prompt}"
         else:
-            user_prompt_text = "Please generate a podcast script based on the uploaded file."
+            user_prompt = "Please generate a podcast script based on the uploaded file."
+
+        # If file is provided, read its content
+        file_content = ""
+        if file_obj:
+            try:
+                file_bytes = await self._read_file_bytes(file_obj)
+                file_content = file_bytes.decode('utf-8', errors='ignore')
+                user_prompt = f"{user_prompt}\n\nFile content:\n{file_content}"
+            except Exception as e:
+                raise Exception(f"Failed to read file: {str(e)}")
 
         messages = [
             {"role": "system", "content": system_prompt},
-            {"role": "user", "content": user_prompt_text}
+            {"role": "user", "content": user_prompt}
         ]
 
-        # افزودن فایل به درخواست (اگر Talkbot.ir API از این فرمت پشتیبانی کند)
-        # توجه: API های LLM معمولا فایل ها را به صورت "base64 encoded" یا "multipart/form-data" 
-        # دریافت می کنند. اگر Talkbot.ir API دقیقا همانند Google Gemini API در اینجا عمل نکند،
-        # این بخش نیاز به تغییر دارد. برای سادگی، فرض می کنیم فقط متن اصلی ارسال می شود
-        # و فایل باید جداگانه پردازش و به متن تبدیل شود، سپس متن آن به مدل ارسال شود.
-        # در این مثال، فرض می کنیم که Gemini Pro از این طریق فایل را دریافت می کند.
-        # اما برای OpenRouter/Deepseek باید فایل را به متن تبدیل کنید.
-        
-        # برای سادگی و تطابق با OpenRouter (معمولا فایل را مستقیم دریافت نمی‌کند)،
-        # اینجا فرض می‌کنیم که اگر فایل بود، محتوایش قبلاً به user_prompt_text اضافه شده.
-        # اگر Talkbot.ir قابلیت پردازش فایل مستقیم را دارد، این بخش را مطابق مستنداتشان تغییر دهید.
-        if file_obj:
-            # این بخش را باید مطابق با نحوه ارسال فایل به Talkbot.ir API تغییر دهید
-            # مثلاً اگر Talkbot API از آپلود فایل پشتیبانی می‌کند، باید از await self._read_file_bytes(file_obj) استفاده کنید.
-            # برای ساده‌سازی فعلی، فرض می‌کنیم فایل‌ها صرفاً برای زمینه (context) هستند
-            # و باید محتوای آن‌ها به صورت متنی به مدل فرستاده شود.
-            # فعلاً، فایل به صورت مستقیم به API چت Deepseek فرستاده نمی‌شود،
-            # مگر اینکه Talkbot.ir API برای Deepseek V3 قابلیت ورودی فایل را داشته باشد.
-            # در صورتی که فایل واقعاً نیاز به پردازش توسط مدل دارد، باید محتوای آن را خوانده و
-            # به string تبدیل کنید و به prompt اضافه کنید.
-            pass # نیاز به پیاده‌سازی تبدیل فایل به متن و اضافه کردن به prompt
-
         headers = {
-            "Authorization": f"Bearer {api_key}",
+            "Authorization": f"Bearer {self.api_key}",
             "Content-Type": "application/json"
         }
 
-        json_payload = {
+        payload = {
             "model": "deepseek-v3-0324",
             "messages": messages,
             "temperature": 1,
-            # "response_mime_type": "application/json" # این پارامتر ممکن است برای Deepseek V3 نباشد
-            # در حالت عادی باید مدل خودش خروجی JSON دهد.
-            # اگر Talkbot.ir برای این مدل پارامتری برای تضمین JSON دارد، اضافه کنید.
+            "response_format": { "type": "json_object" }
         }
 
         try:
             if progress:
                 progress(0.3, "Generating podcast script...")
-                
+            
             async with aiohttp.ClientSession() as session:
-                async with session.post(self.talkbot_openrouter_api_url, headers=headers, json=json_payload, timeout=60) as response:
-                    response.raise_for_status() # برای بررسی خطاهای HTTP
-                    response_data = await response.json()
+                async with session.post(
+                    self.api_url,
+                    headers=headers,
+                    json=payload,
+                    timeout=60
+                ) as response:
+                    
+                    if response.status != 200:
+                        error_msg = await response.text()
+                        raise Exception(f"API request failed: {error_msg}")
+                    
+                    data = await response.json()
+                    response_text = data.get('choices', [{}])[0].get('message', {}).get('content', '')
+                    
+                    if not response_text:
+                        raise Exception("Empty response from API")
+                    
+                    if progress:
+                        progress(0.4, "Script generated successfully!")
+                    
+                    return json.loads(response_text)
                     
-            if not response_data or not response_data.get('choices'):
-                raise Exception("Invalid response from Talkbot.ir API or no choices found.")
-
-            generated_text = response_data['choices'][0]['message']['content']
-            
-            # مدل ممکن است پاسخ را داخل یک JSON block برگرداند (مثل "```json\n...\n```")
-            # باید آن را استخراج کنیم.
-            if generated_text.startswith("```json"):
-                generated_text = generated_text[len("```json"):].strip()
-                if generated_text.endswith("```"):
-                    generated_text = generated_text[:-len("```")].strip()
-
-            print(f"Generated raw text from Deepseek-v3:\n{generated_text}")
-
-            script_json = json.loads(generated_text)
-            
-            if progress:
-                progress(0.4, "Script generated successfully!")
-                
-            return script_json
         except asyncio.TimeoutError:
             raise Exception("The script generation request timed out. Please try again later.")
-        except aiohttp.ClientResponseError as e:
-            if e.status == 401:
-                raise Exception("Unauthorized: Invalid API key or insufficient permissions for Talkbot.ir.")
-            elif e.status == 429:
-                raise Exception("Rate limit exceeded for the API key. Please try again later or provide your own API key.")
-            else:
-                raise Exception(f"Failed to generate podcast script from Talkbot.ir: HTTP Error {e.status} - {e.message}")
-        except json.JSONDecodeError as e:
-            raise Exception(f"Failed to decode JSON from Talkbot.ir response. Raw text: '{generated_text}'. Error: {e}")
+        except json.JSONDecodeError:
+            raise Exception("Invalid JSON response from API")
         except Exception as e:
-            raise Exception(f"Failed to generate podcast script: {e}")
+            if "rate limit" in str(e).lower():
+                raise Exception("Rate limit exceeded. Please try again later.")
+            else:
+                raise Exception(f"Failed to generate podcast script: {e}")
 
-    async def tts_generate(self, text: str, speaker: int, speaker1_type: str, speaker2_type: str, api_key: str) -> str:
-        # speaker1_type و speaker2_type اینجا همان "male" یا "female" هستند
-        voice_type = speaker1_type if speaker == 1 else speaker2_type
-        
-        # پارامتر voice_gender در Talkbot.ir TTS API
-        voice_gender = "male" if voice_type == "male" else "female" 
+    async def _read_file_bytes(self, file_obj) -> bytes:
+        """Read file bytes from a file object"""
+        # Check file size before reading
+        if hasattr(file_obj, 'size'):
+            file_size = file_obj.size
+        else:
+            file_size = os.path.getsize(file_obj.name)
+            
+        if file_size > MAX_FILE_SIZE_BYTES:
+            raise Exception(f"File size exceeds the {MAX_FILE_SIZE_MB}MB limit. Please upload a smaller file.")
+            
+        if hasattr(file_obj, 'read'):
+            return file_obj.read()
+        else:
+            async with aiofiles.open(file_obj.name, 'rb') as f:
+                return await f.read()
+
+    async def tts_generate(self, text: str) -> str:
+        headers = {
+            'accept': 'application/json',
+        }
         
         params = {
-            "text": text,
-            "voice_gender": voice_gender,
-            # اگر Talkbot.ir TTS نیاز به Auth Token دارد:
-            # "api_key": api_key 
-            # یا در هدر Authorization
+            'text': text,
         }
         
-        headers = {
-            # اگر Talkbot.ir TTS نیاز به تایید هویت با Bearer Token دارد
-            "Authorization": f"Bearer {api_key}" 
-        }
-
         temp_filename = f"temp_{uuid.uuid4()}.wav"
         
         try:
             async with aiohttp.ClientSession() as session:
-                # Talkbot.ir/TTS-tkun URL با متد GET و پارامترها
-                async with session.get(self.talkbot_tts_url, params=params, headers=headers, timeout=30) as response:
-                    response.raise_for_status() # برای بررسی خطاهای HTTP
-                    # فرض بر این است که پاسخ مستقیم فایل صوتی با فرمت wav است.
-                    # اگر لینک باشد، باید آن لینک را دانلود کنید.
+                async with session.get(
+                    self.tts_url,
+                    params=params,
+                    headers=headers,
+                    timeout=30
+                ) as response:
                     
-                    # بررسی Content-Type برای اطمینان از اینکه فایل صوتی است
-                    content_type = response.headers.get('Content-Type', '')
-                    if 'audio' not in content_type:
-                        raise Exception(f"Unexpected content type from TTS API: {content_type}. Expected audio.")
-                        
-                    audio_data = await response.read()
+                    if response.status != 200:
+                        error_msg = await response.text()
+                        raise Exception(f"TTS API error: {error_msg}")
                     
-                    if not audio_data:
-                        raise Exception("Received empty audio data from TTS API.")
-
+                    # Save the audio file
                     async with aiofiles.open(temp_filename, 'wb') as f:
-                        await f.write(audio_data)
-                        
-            return temp_filename
+                        await f.write(await response.read())
+                    
+                    return temp_filename
+                    
         except asyncio.TimeoutError:
             if os.path.exists(temp_filename):
                 os.remove(temp_filename)
-            raise Exception("Text-to-speech generation timed out. Please try with a shorter text.")
-        except aiohttp.ClientResponseError as e:
-            if os.path.exists(temp_filename):
-                os.remove(temp_filename)
-            if e.status == 401:
-                raise Exception("Unauthorized access to Talkbot.ir TTS. Invalid API key or token.")
-            else:
-                raise Exception(f"Error from Talkbot.ir TTS API: HTTP {e.status} - {e.message}")
+            raise Exception("Text-to-speech generation timed out.")
         except Exception as e:
             if os.path.exists(temp_filename):
                 os.remove(temp_filename)
-            raise Exception(f"Failed to generate speech with Talkbot.ir TTS: {e}")
+            raise e
 
     async def combine_audio_files(self, audio_files: List[str], progress=None) -> str:
         if progress:
@@ -388,13 +187,8 @@ Follow this example structure:
             
         combined_audio = AudioSegment.empty()
         for audio_file in audio_files:
-            try:
-                combined_audio += AudioSegment.from_file(audio_file)
-            except Exception as e:
-                print(f"Warning: Could not load audio file {audio_file}: {e}")
-            finally:
-                if os.path.exists(audio_file):
-                    os.remove(audio_file)  # Clean up temporary files
+            combined_audio += AudioSegment.from_file(audio_file)
+            os.remove(audio_file)  # Clean up temporary files
 
         output_filename = f"output_{uuid.uuid4()}.wav"
         combined_audio.export(output_filename, format="wav")
@@ -404,13 +198,14 @@ Follow this example structure:
             
         return output_filename
 
-    async def generate_podcast(self, input_text: str, language: str, speaker1_type: str, speaker2_type: str, api_key: str, file_obj=None, progress=None) -> str:
+    async def generate_podcast(self, input_text: str, language: str, file_obj=None, progress=None) -> str:
         try:
             if progress:
                 progress(0.1, "Starting podcast generation...")
                 
+            # Set overall timeout for the entire process
             return await asyncio.wait_for(
-                self._generate_podcast_internal(input_text, language, speaker1_type, speaker2_type, api_key, file_obj, progress),
+                self._generate_podcast_internal(input_text, language, file_obj, progress),
                 timeout=600  # 10 minutes total timeout
             )
         except asyncio.TimeoutError:
@@ -418,26 +213,28 @@ Follow this example structure:
         except Exception as e:
             raise Exception(f"Error generating podcast: {str(e)}")
     
-    async def _generate_podcast_internal(self, input_text: str, language: str, speaker1_type: str, speaker2_type: str, api_key: str, file_obj=None, progress=None) -> str:
+    async def _generate_podcast_internal(self, input_text: str, language: str, file_obj=None, progress=None) -> str:
         if progress:
             progress(0.2, "Generating podcast script...")
             
-        podcast_json = await self.generate_script(input_text, language, api_key, file_obj, progress)
+        podcast_json = await self.generate_script(input_text, language, file_obj, progress)
         
         if progress:
             progress(0.5, "Converting text to speech...")
         
         audio_files = []
         total_lines = len(podcast_json['podcast'])
-        batch_size = 5 # تعداد درخواست های TTS همزمان
         
+        # Process in batches
+        batch_size = 5  # Conservative batch size
         for batch_start in range(0, total_lines, batch_size):
             batch_end = min(batch_start + batch_size, total_lines)
             batch = podcast_json['podcast'][batch_start:batch_end]
             
+            # Create tasks for concurrent processing
             tts_tasks = []
             for item in batch:
-                tts_task = self.tts_generate(item['line'], item['speaker'], speaker1_type, speaker2_type, api_key)
+                tts_task = self.tts_generate(item['line'])
                 tts_tasks.append(tts_task)
             
             try:
@@ -445,20 +242,22 @@ Follow this example structure:
                 
                 for i, result in enumerate(batch_results):
                     if isinstance(result, Exception):
-                        # اگر خطایی در یک TTS بود، همه فایل‌های تولید شده را پاک کنید و خطا دهید
+                        # Clean up any files already created
                         for file in audio_files:
                             if os.path.exists(file):
                                 os.remove(file)
-                        raise Exception(f"Error generating speech for line '{batch[i]['line']}': {str(result)}")
+                        raise Exception(f"Error generating speech: {str(result)}")
                     else:
                         audio_files.append(result)
                         
+                # Update progress
                 if progress:
                     current_progress = 0.5 + (0.4 * (batch_end / total_lines))
                     progress(current_progress, f"Processed {batch_end}/{total_lines} speech segments...")
             
             except Exception as e:
-                for file in audio_files: # اطمینان از پاک شدن فایل‌های موقت در صورت بروز خطا
+                # Clean up any files already created
+                for file in audio_files:
                     if os.path.exists(file):
                         os.remove(file)
                 raise Exception(f"Error in batch TTS generation: {str(e)}")
@@ -466,24 +265,15 @@ Follow this example structure:
         combined_audio = await self.combine_audio_files(audio_files, progress)
         return combined_audio
 
-async def process_input(input_text: str, input_file, language: str, speaker1_type: str, speaker2_type: str, api_key: str = "", progress=None) -> str:
+async def process_input(input_text: str, input_file, language: str, progress=None) -> str:
     start_time = time.time()
 
-    # Talkbot.ir TTS فقط male/female دارد. پس mapping لازم نیست.
-    # speaker1_type و speaker2_type مستقیماً به API فرستاده می‌شوند.
-
     try:
         if progress:
             progress(0.05, "Processing input...")
 
-        if not api_key:
-            # سعی می‌کنیم API key را از متغیر محیطی بخوانیم
-            api_key = os.getenv("TALKBOT_API_KEY") 
-            if not api_key:
-                raise Exception("No API key provided. Please provide your Talkbot.ir/OpenRouter API key.")
-
         podcast_generator = PodcastGenerator()
-        podcast = await podcast_generator.generate_podcast(input_text, language, speaker1_type, speaker2_type, api_key, input_file, progress)
+        podcast = await podcast_generator.generate_podcast(input_text, language, input_file, progress)
 
         end_time = time.time()
         print(f"Total podcast generation time: {end_time - start_time:.2f} seconds")
@@ -492,41 +282,35 @@ async def process_input(input_text: str, input_file, language: str, speaker1_typ
     except Exception as e:
         error_msg = str(e)
         if "rate limit" in error_msg.lower():
-            raise Exception("Rate limit exceeded. Please try again later or use your own API key.")
+            raise Exception("Rate limit exceeded. Please try again later.")
         elif "timeout" in error_msg.lower():
-            raise Exception("The request timed out. This could be due to server load or the length of your input. Please try again with shorter text.")
-        elif "unauthorized" in error_msg.lower() or "api key" in error_msg.lower():
-            raise Exception("Invalid API key. Please check your Talkbot.ir/OpenRouter API key.")
+            raise Exception("The request timed out. Please try again with shorter text.")
         else:
             raise Exception(f"Error: {error_msg}")
 
 # Gradio UI
-def generate_podcast_gradio(input_text, input_file, language, speaker1, speaker2, api_key, progress=gr.Progress()):
+def generate_podcast_gradio(input_text, input_file, language, progress=gr.Progress()):
     # Handle the file if uploaded
     file_obj = None
     if input_file is not None:
         file_obj = input_file
         
+    # Use the progress function from Gradio
     def progress_callback(value, text):
         progress(value, text)
 
-    # Note: Gradio's event handling may automatically manage the asyncio loop.
-    # If not, you might need to run it explicitly using asyncio.run() or similar.
-    # For now, let's assume Gradio handles it for async functions.
-    
-    # برای Gradio، تابعی که به .run() یا .click() متصل می‌شود، می‌تواند async باشد.
-    # Gradio به طور خودکار آن را در یک event loop اجرا می‌کند.
-    return asyncio.run(process_input(
+    # Run the async function in the event loop
+    result = asyncio.run(process_input(
         input_text, 
         file_obj, 
-        language, 
-        speaker1, 
-        speaker2, 
-        api_key,
+        language,
         progress_callback
     ))
+    
+    return result
 
 def main():
+    # Define language options
     language_options = [
         "Auto Detect",
         "Afrikaans", "Albanian", "Amharic", "Arabic", "Armenian", "Azerbaijani",
@@ -544,10 +328,7 @@ def main():
         "Uzbek", "Vietnamese", "Welsh", "Zulu"
     ]
     
-    # Talkbot.ir TTS فقط gender را تعیین می‌کند، نه اسم خاص.
-    # پس اینجا male/female را انتخاب می‌کنیم.
-    voice_options = ["male", "female"]
-    
+    # Create Gradio interface
     with gr.Blocks(title="PodcastGen 🎙️") as demo:
         gr.Markdown("# PodcastGen 🎙️")
         gr.Markdown("Generate a 2-speaker podcast from text input or documents!")
@@ -559,14 +340,7 @@ def main():
             with gr.Column(scale=1):
                 input_file = gr.File(label="Or Upload a PDF or TXT file", file_types=[".pdf", ".txt"])
         
-        with gr.Row():
-            with gr.Column():
-                api_key = gr.Textbox(label="Your Talkbot.ir/OpenRouter API Key (Required)", placeholder="Enter your Talkbot.ir API key here for Gemini Deepseek and TTS.", type="password")
-                language = gr.Dropdown(label="Language", choices=language_options, value="Auto Detect")
-
-            with gr.Column():
-                speaker1 = gr.Dropdown(label="Speaker 1 Voice (Gender)", choices=voice_options, value="male")
-                speaker2 = gr.Dropdown(label="Speaker 2 Voice (Gender)", choices=voice_options, value="female")
+        language = gr.Dropdown(label="Language", choices=language_options, value="Auto Detect")
         
         generate_btn = gr.Button("Generate Podcast", variant="primary")
         
@@ -575,12 +349,11 @@ def main():
             
         generate_btn.click(
             fn=generate_podcast_gradio,
-            inputs=[input_text, input_file, language, speaker1, speaker2, api_key],
+            inputs=[input_text, input_file, language],
             outputs=[output_audio]
         )
     
-    demo.launch(share=False) # share=True برای اشتراک گذاری لینک عمومی
+    demo.launch()
 
 if __name__ == "__main__":
     main()
-