Add MOSS-SoundEffect generation (CLI + webui section)

New tts/generate_sfx.py mirrors generate.py (MOSS-TTS) but for the
MOSS-SoundEffect checkpoint: text-only ambience/SFX generation, no
reference audio or voice cloning. Needed PYTORCH_ALLOC_CONF=
expandable_segments:True baked in - this checkpoints weights alone use
~22.3GB of this boxs ~23.56GB usable VRAM, leaving generate() itself no
room without it (confirmed via OOM otherwise).

webui/app.py gets a new standalone "5. Sound effects" section in the
Pipeline Studio tab (separate from the AuK/MOSS-TTS voice flow - this is
a different generation primitive with no reference/history state).
Verified end-to-end: three real test clips (door creak, market ambience,
car pass) generated cleanly, approved by Steve as "definitely passable,"
then wired into the GUI and confirmed working through app.py itself.

Co-Authored-By: Claude Sonnet 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01SN15Vxn4svKuhYxFsVtMJf

webui/app.py

@@ -49,6 +49,7 @@ AUK_FLASH_CKPT = AUK_DIR / "ckpts" / "AuK-Flash" / "auk_flash.safetensors"
 AUK_QWEN_PATH = AUK_DIR / "ckpts" / "Qwen2.5-Omni-3B"
 TTS_PYTHON = TTS_DIR / ".venv" / "bin" / "python3"
 TTS_GENERATE = TTS_DIR / "generate.py"
+TTS_GENERATE_SFX = TTS_DIR / "generate_sfx.py"
 
 MODEL_BASE = "Base (full quality)"
 MODEL_FLASH = "Flash (fast draft, 4-step)"
@@ -418,6 +419,30 @@ def generate_speech(reference: str, text: str, language: str):
     return str(out_path), transcript
 
 
+def generate_sound_effect(description: str):
+    """MOSS-SoundEffect - text-only ambience/SFX generation, no reference
+    audio or character voice involved (different generation primitive from
+    the AuK/MOSS-TTS voice-cloning flow above, so it's a standalone section
+    rather than wired into that pipeline's history/reference state).
+    No whisper verification here - there's no speech content to check."""
+    if not description or not description.strip():
+        raise gr.Error("Enter a description of the sound effect/ambience.")
+
+    _free_comfyui_memory()
+    out_path = WORKDIR / f"sfx_{int(time.time())}.wav"
+    env = None
+    if TTS_LD_LIBRARY_PATH:
+        env = os.environ.copy()
+        env["LD_LIBRARY_PATH"] = TTS_LD_LIBRARY_PATH
+
+    _run(
+        [str(TTS_PYTHON), str(TTS_GENERATE_SFX), "--text", description.strip(), "--output", str(out_path)],
+        cwd=TTS_DIR,
+        env=env,
+    )
+    return str(out_path)
+
+
 # ---------------------------------------------------------------------------
 # Characters (the-service-station cast)
 # ---------------------------------------------------------------------------
@@ -781,6 +806,24 @@ with gr.Blocks(title="Audio-demo pipeline studio") as demo:
                     save_char_btn = gr.Button("Save as new character (status: draft)")
                     save_char_status = gr.Markdown()
 
+            with gr.Row():
+                with gr.Column():
+                    gr.Markdown(
+                        "## 5. Sound effects (MOSS-SoundEffect)\n"
+                        "Text-only ambience/SFX generation - no reference audio or character "
+                        "voice involved, a separate model from the voice pipeline above."
+                    )
+                    sfx_description = gr.Textbox(
+                        label="Description",
+                        placeholder=(
+                            'e.g. "A door creaking open slowly, then closing with a soft thud." '
+                            'or "A busy outdoor market with distant chatter and footsteps."'
+                        ),
+                        lines=2,
+                    )
+                    sfx_btn = gr.Button("Generate sound effect", variant="primary")
+                    sfx_audio = gr.Audio(label="Generated sound effect", type="filepath")
+
             strip_btn.click(strip_vocals, inputs=[input_audio], outputs=[vocals_out, no_vocals_out])
             use_vocals_btn.click(
                 use_as_reference, inputs=[vocals_out, history_state],
@@ -819,6 +862,7 @@ with gr.Blocks(title="Audio-demo pipeline studio") as demo:
                 inputs=[new_char_name, new_char_role, new_char_prose, gen_language, current_ref_audio, history_state],
                 outputs=[save_char_status],
             )
+            sfx_btn.click(generate_sound_effect, inputs=[sfx_description], outputs=[sfx_audio])
 
         with gr.Tab("Characters"):
             gr.Markdown(