refactor: integrate voice selection utility in podcast audio generation

- remove NavUser component and update sidebar layout;

surfsense_backend/app/agents/podcaster/nodes.py

@@ -16,6 +16,7 @@ from app.services.llm_service import get_user_long_context_llm
 from .configuration import Configuration
 from .prompts import get_podcast_generation_prompt
 from .state import PodcastTranscriptEntry, PodcastTranscripts, State
+from .utils import get_voice_for_provider
 
 
 async def create_podcast_transcript(
@@ -121,16 +122,6 @@ async def create_merged_podcast_audio(
     output_path = f"podcasts/{session_id}_podcast.mp3"
     os.makedirs("podcasts", exist_ok=True)
 
-    # Map of speaker_id to voice
-    voice_mapping = {
-        0: "alloy",  # Default/intro voice
-        1: "echo",  # First speaker
-        # 2: "fable",  # Second speaker
-        # 3: "onyx",   # Third speaker
-        # 4: "nova",   # Fourth speaker
-        # 5: "shimmer" # Fifth speaker
-    }
-
     # Generate audio for each transcript segment
     audio_files = []
 
@@ -144,7 +135,7 @@ async def create_merged_podcast_audio(
             dialog = segment.get("dialog", "")
 
         # Select voice based on speaker_id
-        voice = voice_mapping.get(speaker_id, "alloy")
+        voice = get_voice_for_provider(app_config.TTS_SERVICE, speaker_id)
 
         # Generate a unique filename for this segment
         filename = f"{temp_dir}/{session_id}_{index}.mp3"

surfsense_backend/app/agents/podcaster/utils.py

@@ -0,0 +1,69 @@
+def get_voice_for_provider(provider: str, speaker_id: int) -> dict | str:
+    """
+    Get the appropriate voice configuration based on the TTS provider and speaker ID.
+
+    Args:
+        provider: The TTS provider (e.g., "openai/tts-1", "vertex_ai/test")
+        speaker_id: The ID of the speaker (0-5)
+
+    Returns:
+        Voice configuration - string for OpenAI, dict for Vertex AI
+    """
+    # Extract provider type from the model string
+    provider_type = (
+        provider.split("/")[0].lower() if "/" in provider else provider.lower()
+    )
+
+    if provider_type == "openai":
+        # OpenAI voice mapping - simple string values
+        openai_voices = {
+            0: "alloy",  # Default/intro voice
+            1: "echo",  # First speaker
+            2: "fable",  # Second speaker
+            3: "onyx",  # Third speaker
+            4: "nova",  # Fourth speaker
+            5: "shimmer",  # Fifth speaker
+        }
+        return openai_voices.get(speaker_id, "alloy")
+
+    elif provider_type == "vertex_ai":
+        # Vertex AI voice mapping - dict with languageCode and name
+        vertex_voices = {
+            0: {
+                "languageCode": "en-US",
+                "name": "en-US-Studio-O",
+            },
+            1: {
+                "languageCode": "en-US",
+                "name": "en-US-Studio-M",
+            },
+            2: {
+                "languageCode": "en-UK",
+                "name": "en-UK-Studio-A",
+            },
+            3: {
+                "languageCode": "en-UK",
+                "name": "en-UK-Studio-B",
+            },
+            4: {
+                "languageCode": "en-AU",
+                "name": "en-AU-Studio-A",
+            },
+            5: {
+                "languageCode": "en-AU",
+                "name": "en-AU-Studio-B",
+            },
+        }
+        return vertex_voices.get(speaker_id, vertex_voices[0])
+
+    else:
+        # Default fallback to OpenAI format for unknown providers
+        default_voices = {
+            0: "alloy",
+            1: "echo",
+            2: "fable",
+            3: "onyx",
+            4: "nova",
+            5: "shimmer",
+        }
+        return default_voices.get(speaker_id, "alloy")