diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..078fa2c
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,40 @@
+name: CI
+
+on:
+  push:
+    branches: [master, main]
+  pull_request:
+
+jobs:
+  python:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+      - name: Install
+        run: python -m pip install -e '.[dev]'
+      - name: Test
+        run: pytest -q
+      - name: Lint
+        run: ruff check .
+
+  frontend:
+    runs-on: ubuntu-latest
+    defaults:
+      run:
+        working-directory: frontend
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: "22"
+          cache: npm
+          cache-dependency-path: frontend/package-lock.json
+      - name: Install
+        run: npm ci
+      - name: Lint
+        run: npm run lint
+      - name: Build
+        run: npm run build
diff --git a/README.md b/README.md
index 8c5d020..43918ec 100644
--- a/README.md
+++ b/README.md
@@ -48,6 +48,16 @@ Input:                          Output:
 - **CLI + Web UI** — Use from terminal or browser
 - **Cloud GPU Ready** — Runs on Google Colab (free T4) or any CUDA server
 
+## Choose Your Run Mode
+
+| Mode | Best For | Command |
+| --- | --- | --- |
+| Gradio local | Quick single-process demo | `python app.py` |
+| Gradio remote TTS | Using Kaggle/Colab GPU TTS | `TTS_BASE_URL=https://your-ngrok-url.ngrok-free.dev python app_remote.py` |
+| FastAPI backend | API server for the Next.js frontend | `python -m backend.main` |
+| Next.js frontend | Full web UI | `cd frontend && npm run dev` |
+| CLI | Script-to-WAV generation | `podforge --script examples/demo_drama.txt --output output.wav --tts-url http://localhost:8809` |
+
 ## Quick Start
 
 ### Deploy Your Own Space
@@ -153,6 +163,39 @@ cd frontend && npm install && npm run dev
 
 Open `http://localhost:3000` → load a demo script → pick voices → click **生成播客**.
 
+## Verification
+
+Run the Python checks from the repository root:
+
+```bash
+pip install -e '.[dev]'
+pytest -q
+ruff check .
+```
+
+Run the frontend checks from `frontend/`:
+
+```bash
+npm ci
+npm run lint
+npm run build
+```
+
+### Dependency Audit Note
+
+`npm audit` may report a moderate `postcss` advisory through Next.js. Do not run `npm audit fix --force` unless Next.js provides a safe non-breaking upgrade path; the current forced fix proposes a destructive downgrade to `next@9.3.3`. Treat this as a monitored dependency risk, not an immediate force-fix task.
+
+### Manual UI Regression Check
+
+This check does not require a live TTS server:
+
+1. Start the frontend with `cd frontend && npm run dev`.
+2. Open `http://localhost:3000`.
+3. Click **加载示例**.
+4. Confirm the editor footer and voice panel show parsed script content.
+5. Select all script text and delete it.
+6. Confirm the editor footer shows `0` roles and `0` lines, and the voice panel returns to its empty state.
+
 ## Script Format
 
 ```text
diff --git a/backend/tts_client.py b/backend/tts_client.py
index a05b822..2b34737 100644
--- a/backend/tts_client.py
+++ b/backend/tts_client.py
@@ -22,14 +22,14 @@ class VoxCPMClient:
     def __init__(
         self,
         base_url: str | None = None,
-        timeout: float = 120.0,
+        timeout: float | None = None,
     ):
         self.base_url = (
             base_url
             or os.environ.get("PODFORGE_TTS_URL")
             or "http://localhost:8809"
         ).rstrip("/")
-        self.timeout = timeout
+        self.timeout = timeout if timeout is not None else float(os.environ.get("PODFORGE_TTS_TIMEOUT", "300"))
         self._client: httpx.AsyncClient | None = None
 
     @property
diff --git a/frontend/app/page.tsx b/frontend/app/page.tsx
index 98a2b22..9cfdf92 100644
--- a/frontend/app/page.tsx
+++ b/frontend/app/page.tsx
@@ -196,7 +196,7 @@ export default function Home() {
           <div className="flex-1 min-h-0 overflow-auto mb-4">
             <VoicePanel
               presets={presets}
-              lines={parsedLines}
+              lines={lines}
               voiceOverrides={voiceOverrides}
               onOverride={(char, desc) =>
                 setVoiceOverrides((prev) => ({ ...prev, [char]: desc }))
@@ -209,7 +209,7 @@ export default function Home() {
                 {error}
               </div>
             )}
-            {generating ? (
+            {generating || (error && progress.current > 0) ? (
               <ProgressTracker
                 current={progress.current}
                 total={progress.total}
diff --git a/pyproject.toml b/pyproject.toml
index 4ccbe52..3516025 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -9,6 +9,7 @@ dependencies = [
     "fastapi>=0.115.0",
     "uvicorn[standard]>=0.30.0",
     "httpx>=0.27.0",
+    "socksio>=1.0.0",
     "pydantic>=2.0",
     "soundfile>=0.12.0",
     "numpy>=1.26.0",
diff --git a/requirements.txt b/requirements.txt
index 0fa6661..b8c5275 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,6 +1,7 @@
 fastapi>=0.115.0
 uvicorn[standard]>=0.30.0
 httpx>=0.27.0
+socksio>=1.0.0
 pydantic>=2.0
 soundfile>=0.12.0
 numpy>=1.26.0