Update app.py

app.py

@@ -9,7 +9,45 @@ from typing import Optional
 SUPPORTED_FORMATS = ['.mp3', '.wav', '.m4a', '.aac', '.ogg', '.flac', '.webm']
 MAX_FILE_SIZE_MB = 25
 
-def transcribe_with_groq(audio_file, api_key, enable_diarization=False):
+# Language options for Whisper
+LANGUAGE_OPTIONS = [
+    ("🌍 Auto-rilevamento", "auto"),
+    ("🇮🇹 Italiano", "it"),
+    ("🇺🇸 English", "en"),
+    ("🇪🇸 Español", "es"),
+    ("🇫🇷 Français", "fr"),
+    ("🇩🇪 Deutsch", "de"),
+    ("🇵🇹 Português", "pt"),
+    ("🇷🇺 Русский", "ru"),
+    ("🇨🇳 中文", "zh"),
+    ("🇯🇵 日本語", "ja"),
+    ("🇰🇷 한국어", "ko"),
+    ("🇳🇱 Nederlands", "nl"),
+    ("🇸🇦 العربية", "ar"),
+    ("🇮🇳 हिन्दी", "hi"),
+    ("🇹🇷 Türkçe", "tr"),
+    ("🇵🇱 Polski", "pl"),
+    ("🇸🇪 Svenska", "sv"),
+    ("🇳🇴 Norsk", "no"),
+    ("🇩🇰 Dansk", "da"),
+    ("🇫🇮 Suomi", "fi"),
+    ("🇬🇷 Ελληνικά", "el"),
+    ("🇮🇱 עברית", "he"),
+    ("🇹🇭 ไทย", "th"),
+    ("🇻🇳 Tiếng Việt", "vi"),
+    ("🇮🇩 Bahasa Indonesia", "id"),
+    ("🇲🇾 Bahasa Melayu", "ms"),
+    ("🇺🇦 Українська", "uk"),
+    ("🇨🇿 Čeština", "cs"),
+    ("🇭🇺 Magyar", "hu"),
+    ("🇷🇴 Română", "ro"),
+    ("🇧🇬 Български", "bg"),
+    ("🇭🇷 Hrvatski", "hr"),
+    ("🇸🇰 Slovenčina", "sk"),
+    ("🇸🇮 Slovenščina", "sl"),
+]
+
+def transcribe_with_groq(audio_file, api_key, enable_diarization=False, language="auto"):
     """Transcribe using Groq's free Whisper API with basic speaker detection"""
     try:
         url = "https://api.groq.com/openai/v1/audio/transcriptions"
@@ -21,6 +59,10 @@ def transcribe_with_groq(audio_file, api_key, enable_diarization=False):
                 "timestamp_granularities[]": (None, "segment")
             }
             
+            # Add language parameter if not auto
+            if language != "auto":
+                data["language"] = (None, language)
+            
             files = {"file": f}
             headers = {"Authorization": f"Bearer {api_key}"}
             
@@ -75,7 +117,7 @@ def format_with_speakers_groq(segments):
     
     return formatted_text.strip()
 
-def transcribe_with_assemblyai(audio_file, api_key, enable_diarization=False):
+def transcribe_with_assemblyai(audio_file, api_key, enable_diarization=False, language="auto"):
     """Transcribe using AssemblyAI's API with advanced speaker diarization"""
     try:
         # Step 1: Upload file
@@ -89,13 +131,19 @@ def transcribe_with_assemblyai(audio_file, api_key, enable_diarization=False):
         
         upload_url = response.json()["upload_url"]
         
-        # Step 2: Request transcription with diarization
+        # Step 2: Request transcription with diarization and language
         data = {
             "audio_url": upload_url,
             "speaker_labels": enable_diarization,
             "speakers_expected": 2 if enable_diarization else None
         }
         
+        # Add language detection or specific language
+        if language == "auto":
+            data["language_detection"] = True
+        else:
+            data["language_code"] = language
+        
         response = requests.post("https://api.assemblyai.com/v2/transcript", json=data, headers=headers)
         
         if response.status_code != 200:
@@ -160,21 +208,26 @@ def format_with_speakers_assemblyai(utterances):
     
     return formatted_text.strip()
 
-def transcribe_with_openai(audio_file, api_key):
+def transcribe_with_openai(audio_file, api_key, language="auto"):
     """Transcribe using OpenAI's Whisper API (no diarization support)"""
     try:
         url = "https://api.openai.com/v1/audio/transcriptions"
         
         with open(audio_file, "rb") as f:
-            files = {
-                "file": f,
+            data = {
                 "model": (None, "whisper-1"),
             }
+            
+            # Add language parameter if not auto
+            if language != "auto":
+                data["language"] = (None, language)
+            
+            files = {"file": f}
             headers = {
                 "Authorization": f"Bearer {api_key}"
             }
             
-            response = requests.post(url, files=files, headers=headers, timeout=60)
+            response = requests.post(url, files=files, data=data, headers=headers, timeout=60)
             
         if response.status_code == 200:
             result = response.json()
@@ -210,7 +263,7 @@ def transcribe_with_deepgram(audio_file, api_key):
     except Exception as e:
         return {"success": False, "error": f"Deepgram request failed: {str(e)}"}
 
-def transcribe_audio(audio_file, service_choice, api_key, enable_diarization, progress=gr.Progress()):
+def transcribe_audio(audio_file, service_choice, api_key, enable_diarization, language_choice, progress=gr.Progress()):
     """Main transcription function for audio files only"""
     
     if audio_file is None:
@@ -267,15 +320,15 @@ def transcribe_audio(audio_file, service_choice, api_key, enable_diarization, pr
         # Choose transcription service
         if service_choice == "Groq (Gratis e Veloce)":
             progress(0.3, desc="Trascrivendo con Groq...")
-            result = transcribe_with_groq(audio_file, api_key, enable_diarization)
+            result = transcribe_with_groq(audio_file, api_key, enable_diarization, language_choice)
         elif service_choice == "AssemblyAI (Tier Gratuito)":
             progress(0.3, desc="Caricando su AssemblyAI...")
-            result = transcribe_with_assemblyai(audio_file, api_key, enable_diarization)
+            result = transcribe_with_assemblyai(audio_file, api_key, enable_diarization, language_choice)
         elif service_choice == "OpenAI Whisper (A Pagamento)":
             if enable_diarization:
                 return "❌ OpenAI Whisper non supporta il riconoscimento interlocutori. Usa Groq o AssemblyAI per questa funzione.", ""
             progress(0.3, desc="Trascrivendo con OpenAI...")
-            result = transcribe_with_openai(audio_file, api_key)
+            result = transcribe_with_openai(audio_file, api_key, language_choice)
         elif service_choice == "Deepgram (Prova Gratuita)":
             if enable_diarization:
                 return "❌ Riconoscimento interlocutori non implementato per Deepgram. Usa Groq o AssemblyAI.", ""
@@ -299,6 +352,7 @@ def transcribe_audio(audio_file, service_choice, api_key, enable_diarization, pr
         
         # Create summary
         diarization_status = "✅ Attivo" if enable_diarization else "❌ Disattivo"
+        language_display = "🌍 Auto-rilevamento" if language_choice == "auto" else f"🗣️ {language_choice.upper()}"
         
         summary = f"""✅ **Trascrizione Completata!**
 
@@ -312,6 +366,7 @@ def transcribe_audio(audio_file, service_choice, api_key, enable_diarization, pr
 🔧 **Impostazioni:**
 • Chiave API: {"✅ Fornita" if api_key else "❌ Mancante"}
 • Riconoscimento Interlocutori: {diarization_status}
+• Lingua: {language_display}
 • Formato: {file_extension.upper()}
 """
         
@@ -420,6 +475,14 @@ with gr.Blocks(title=title, theme=gr.themes.Soft(), css=css) as demo:
                     info="Identifica diversi parlanti come Interlocutore 1, 2, ecc."
                 )
                 
+                # Language selection
+                language_dropdown = gr.Dropdown(
+                    choices=LANGUAGE_OPTIONS,
+                    value="auto",
+                    label="🌍 Lingua Audio",
+                    info="Seleziona la lingua dell'audio per migliorare l'accuratezza"
+                )
+                
                 with gr.Accordion("ℹ️ Come ottenere le chiavi API", open=False):
                     gr.Markdown(get_service_info())
             
@@ -452,6 +515,11 @@ with gr.Blocks(title=title, theme=gr.themes.Soft(), css=css) as demo:
 • ✅ **AssemblyAI**: Rilevamento avanzato (AI)
 • ❌ **OpenAI/Deepgram**: Non supportato
 
+**🌍 Selezione Lingua:**
+• **Auto-rilevamento**: Lascia che l'AI detecti la lingua
+• **Lingua specifica**: Migliora accuratezza e velocità
+• **35+ lingue supportate**: Italiano, English, Español, etc.
+
 **Formati supportati:**
 MP3, WAV, M4A, AAC, OGG, FLAC, WebM
                 """)
@@ -477,7 +545,7 @@ MP3, WAV, M4A, AAC, OGG, FLAC, WebM
     # Event handlers
     transcribe_btn.click(
         fn=transcribe_audio,
-        inputs=[audio_input, service_choice, api_key_input, diarization_checkbox],
+        inputs=[audio_input, service_choice, api_key_input, diarization_checkbox, language_dropdown],
         outputs=[output_text, summary_output]
     )
     
@@ -510,7 +578,7 @@ MP3, WAV, M4A, AAC, OGG, FLAC, WebM
     **🔗 Ottieni Chiavi API:**
     [Groq Console](https://console.groq.com/) • [AssemblyAI](https://www.assemblyai.com/) • [OpenAI Platform](https://platform.openai.com/) • [Deepgram](https://deepgram.com/)
     
-    **📚 Info:** Usa API affidabili per trascrizione • Groq e AssemblyAI supportano riconoscimento interlocutori • Tutti i servizi usano modelli Whisper
+    **📚 Info:** Usa API affidabili per trascrizione • Groq e AssemblyAI supportano riconoscimento interlocutori • Supporto per 35+ lingue • Tutti i servizi usano modelli Whisper
     """)
 
 if __name__ == "__main__":