Update src/ui/state.py

src/ui/state.py

@@ -2,15 +2,12 @@
 ui/state.py
 ──────────────────────────────────────────────────────────────────────────────
 VoiceVerse Pro — Centralised Session State & Config Contracts
-
-All UI modules share these typed dataclasses instead of reading
-st.session_state keys directly. This makes every stage independently
-testable and the data-flow explicit.
 """
 
 from __future__ import annotations
 
 from dataclasses import dataclass, field
+from enum import Enum
 from typing import List, Optional
 
 from modules import RetrievedContext, IngestedFile
@@ -18,7 +15,16 @@ from modules.tts_engine import TTSBackend
 
 
 # ──────────────────────────────────────────────────────────────────────────────
-# Sidebar configuration — produced by ui/sidebar.py, consumed by all stages
+# Output mode
+# ──────────────────────────────────────────────────────────────────────────────
+
+class OutputMode(str, Enum):
+    TRANSCRIPT = "Audio Transcript"     # single narrator voice
+    PODCAST    = "Podcast (2 Speakers)" # HOST (female) + GUEST (male) dialogue
+
+
+# ──────────────────────────────────────────────────────────────────────────────
+# Sidebar configuration
 # ──────────────────────────────────────────────────────────────────────────────
 
 @dataclass
@@ -26,6 +32,9 @@ class SidebarConfig:
     # Auth
     hf_token: str = ""
 
+    # Mode
+    output_mode: OutputMode = OutputMode.TRANSCRIPT
+
     # RAG
     top_k: int = 4
     chunk_size: int = 1000
@@ -39,11 +48,15 @@ class SidebarConfig:
 
     # TTS
     tts_backend: TTSBackend = TTSBackend.SPEECHT5
+    # Single-speaker (transcript mode)
     speaker_id: int = 7306
+    # Dual-speaker (podcast mode) — CMU Arctic xvectors
+    female_speaker_id: int = 1580   # SLT-style female
+    male_speaker_id: int = 7306     # BDL-style male
 
 
 # ──────────────────────────────────────────────────────────────────────────────
-# Pipeline state — persisted in st.session_state, mutated by each stage
+# Pipeline state
 # ──────────────────────────────────────────────────────────────────────────────
 
 @dataclass
@@ -60,68 +73,41 @@ class PipelineState:
     """
     stage: int = 0
 
-    # Stage ① outputs
     ingested_files: List[IngestedFile] = field(default_factory=list)
     total_chunks: int = 0
-
-    # Stage ② outputs
     retrieved_context: Optional[RetrievedContext] = None
-
-    # Stage ③ outputs
     generated_script: Optional[str] = None
-
-    # Stage ④ outputs
     audio_bytes: Optional[bytes] = None
     audio_format: str = "audio/wav"
 
-    # ── Helpers ───────────────────────────────────────────────────────────────
-
     def reset_from(self, stage: int) -> None:
-        """Clear all state at and after `stage`."""
         if stage <= 1:
-            self.ingested_files = []
-            self.total_chunks = 0
-            self.stage = 0
+            self.ingested_files = []; self.total_chunks = 0; self.stage = 0
         if stage <= 2:
             self.retrieved_context = None
-            if self.stage >= 2:
-                self.stage = 1
+            if self.stage >= 2: self.stage = 1
         if stage <= 3:
             self.generated_script = None
-            if self.stage >= 3:
-                self.stage = 2
+            if self.stage >= 3: self.stage = 2
         if stage <= 4:
-            self.audio_bytes = None
-            self.audio_format = "audio/wav"
-            if self.stage >= 4:
-                self.stage = 3
+            self.audio_bytes = None; self.audio_format = "audio/wav"
+            if self.stage >= 4: self.stage = 3
 
     @property
-    def has_index(self) -> bool:
-        return self.stage >= 1
-
+    def has_index(self) -> bool: return self.stage >= 1
     @property
-    def has_context(self) -> bool:
-        return self.stage >= 2 and self.retrieved_context is not None
-
+    def has_context(self) -> bool: return self.stage >= 2 and self.retrieved_context is not None
     @property
-    def has_script(self) -> bool:
-        return self.stage >= 3 and self.generated_script is not None
-
+    def has_script(self) -> bool: return self.stage >= 3 and self.generated_script is not None
     @property
-    def has_audio(self) -> bool:
-        return self.stage >= 4 and self.audio_bytes is not None
+    def has_audio(self) -> bool: return self.stage >= 4 and self.audio_bytes is not None
 
 
 # ──────────────────────────────────────────────────────────────────────────────
-# Session state bootstrap (call once in app.py)
+# Session bootstrap
 # ──────────────────────────────────────────────────────────────────────────────
 
 def get_pipeline_state() -> PipelineState:
-    """
-    Retrieve or initialise PipelineState from st.session_state.
-    Always returns the same object within a session.
-    """
     import streamlit as st
     if "pipeline_state" not in st.session_state:
         st.session_state["pipeline_state"] = PipelineState()