Add app.py and requirements

app.py

@@ -0,0 +1,215 @@
+import gradio as gr
+import requests
+import os
+
+VM_ENDPOINT = os.environ.get("VM_ENDPOINT", "")  # Add later via HF Secrets
+
+# ── Your samples + transcripts ──────────────────────────────
+SAMPLES = [
+    {
+        "transcript": "Hello, this is Priya from customer support. [clears throat] I’m calling regarding your refund request. The amount has been processed and should appear in your account within three business days.",
+        "file": "samples/sample_01.wav"
+    },
+    {
+        "transcript": "Hey Rahul, are we still meeting for dinner tonight? [chuckles] I actually reached early for once, [m_laugh] so don’t make me wait outside again.",
+        "file": "samples/sample_02.wav"
+    },
+    {
+        "transcript": "Good morning. [m_clear throat] You have a meeting with the research team at 10 A.M., followed by a dentist appointment in the evening.",
+        "file": "samples/sample_03.wav"
+    },
+    {
+        "transcript": "The current temperature is twenty-seven degrees Celsius. [clears throat] Light rain is expected after 4 P.M., so carrying an umbrella is recommended.",
+        "file": "samples/sample_04.wav"
+    },
+    {
+        "transcript": "Did you finish your homework before watching videos? [sigh] Dinner is almost ready, and I’ve reminded you several times already.",
+        "file": "samples/sample_05.wav"
+    },
+    {
+        "transcript": "Your order is currently being prepared and should arrive soon. [m_clear throat] Thank you for choosing our delivery service today.",
+        "file": "samples/sample_06.wav"
+    },
+    {
+        "transcript": "Hello, this is a reminder for your appointment tomorrow at 11:30 A.M. [clears throat] Please arrive ten minutes early for registration.",
+        "file": "samples/sample_07.wav"
+    },
+    {
+        "transcript": "We need to finalize the evaluation report before Friday. [m_sigh] Can you also review the benchmark results after lunch?",
+        "file": "samples/sample_08.wav"
+    },
+    {
+        "transcript": "Attention passengers traveling to Bengaluru. [clears throat] Your flight has been delayed due to weather conditions. We regret the inconvenience.",
+        "file": "samples/sample_09.wav"
+    },
+    {
+        "transcript": "This is an automated security notification. [m_clear throat] A login attempt was detected from a new device associated with your account.",
+        "file": "samples/sample_10.wav"
+    },
+    {
+        "transcript": "Hey, you sounded stressed yesterday. [m_sigh] I just wanted to know if you’re feeling any better today.",
+        "file": "samples/sample_11.wav"
+    },
+    {
+        "transcript": "The living room lights are now turned off. [clears throat] Your washing machine cycle will finish in approximately twelve minutes.",
+        "file": "samples/sample_12.wav"
+    },
+    {
+        "transcript": "Please submit your assignments before midnight today. [sigh] Late submissions may receive reduced marks without prior approval.",
+        "file": "samples/sample_13.wav"
+    },
+    {
+        "transcript": "Hi, I ordered a laptop last week and it still hasn’t arrived. [sigh] I need an update because this delay is affecting my work.",
+        "file": "samples/sample_14.wav"
+    },
+    {
+        "transcript": "Warning. Heavy rainfall may cause flooding in nearby areas tonight. [m_clear throat] Residents are advised to avoid unnecessary travel.",
+        "file": "samples/sample_15.wav"
+    },
+    {
+        "transcript": "We should leave early on Saturday morning to avoid traffic. [s_laugh] Last time we spent four hours stuck on the highway.",
+        "file": "samples/sample_16.wav"
+    },
+    {
+        "transcript": "Hello, I’m calling regarding your interview for the machine learning engineer role. [clears throat] We’d like to schedule a technical discussion this week.",
+        "file": "samples/sample_17.wav"
+    },
+    {
+        "transcript": "I found three products matching your search query. [m_clear throat] The highest-rated option currently has a fifteen percent discount.",
+        "file": "samples/sample_18.wav"
+    },
+    {
+        "transcript": "[m_laugh] When I was your age, we walked to school every day. Even during heavy rain, we carried books in plastic covers.",
+        "file": "samples/sample_19.wav"
+    },
+    {
+        "transcript": "Did you watch the new AI demo yesterday? [chuckles] The response speed looked impressive, [m_sigh] though a few answers were inaccurate.",
+        "file": "samples/sample_20.wav"
+    }
+]
+
+SSML_TAGS = """
+| Tag | Effect |  Example Usage|
+|-----|--------|----------------|
+| `[clear_throat]` | Clear throat — highest intensity | `[clear_throat] Hello, welcome.` |
+| `[m_clear_throat]` | Clear throat — medium intensity | `[m_clear_throat] Let me explain.` |
+| `[sighs]` | Sigh — highest intensity | `I can't believe it. [sighs]` |
+| `[m_sighs]` | Sigh — medium intensity | `[m_sighs] That was a long day.` |
+| `[laugh]` | Laugh — highest intensity | `That's so funny! [laugh]` |
+| `[s_laugh]` | Laugh — subtle/lesser intensity | `Oh really? [s_laugh] Interesting.` |
+| `[chuckles]` | Chuckles — highest intensity | `Well [chuckles] that was unexpected.` |
+"""
+
+# ── Inference function ───────────────────────────────────────
+def run_tts(text):
+    if not text.strip():
+        return None, "⚠️ Please enter some text."
+    if len(text) > 500:
+        return None, "⚠️ Please keep input under 500 characters."
+    if not VM_ENDPOINT:
+        return None, "⚠️ Live inference endpoint not configured yet."
+    try:
+        response = requests.post(
+            VM_ENDPOINT,
+            json={"text": text},
+            timeout=30
+        )
+        if response.status_code == 200:
+            tmp_path = "/tmp/tts_output.wav"
+            with open(tmp_path, "wb") as f:
+                f.write(response.content)
+            return tmp_path, "✅ Audio generated!"
+        else:
+            return None, f"❌ Server error: {response.status_code}"
+    except requests.exceptions.Timeout:
+        return None, "❌ Request timed out."
+    except Exception as e:
+        return None, f"❌ Error: {str(e)}"
+
+# ── UI ────────────────────────────────────────────────────────
+with gr.Blocks(
+    title="TTS Demo",
+    theme=gr.themes.Soft(),
+    css="""
+        .sample-row { border: 1px solid #e0e0e0; border-radius: 8px; padding: 12px; margin-bottom: 8px; }
+        .ssml-table { font-size: 0.85em; }
+        footer { display: none !important; }
+    """
+) as demo:
+
+    gr.Markdown("""
+    # 🗣️ TTS Demo
+    **Single female voice · English · SSML supported**
+     |  [📄 Paper](#)  |  [💻 Code](#)
+    """)
+
+    # ── Top Section: Input | Output ──────────────────────────
+    with gr.Row(equal_height=True):
+
+        # Left — Text Input
+        with gr.Column(scale=1):
+            gr.Markdown("### ✍️ Input Text")
+            text_input = gr.Textbox(
+                label="",
+                placeholder="Type your text here...\nYou can use SSML tags from the reference below.",
+                lines=8,
+                max_lines=12,
+            )
+
+            # SSML reference below the text box
+            gr.Markdown("#### 🏷️ Supported SSML Tags")
+            gr.Markdown(SSML_TAGS, elem_classes=["ssml-table"])
+
+            with gr.Row():
+                submit_btn = gr.Button("🔊 Generate Audio", variant="primary", scale=2)
+                clear_btn = gr.Button("🗑️ Clear", scale=1)
+
+            status_msg = gr.Markdown("")
+
+        # Right — Audio Output
+        with gr.Column(scale=1):
+            gr.Markdown("### 🎧 Generated Audio")
+            audio_output = gr.Audio(
+                label="",
+                interactive=False,
+                show_download_button=True,
+            )
+            gr.Markdown("""
+            <br><br>
+            **Tips:**
+            - Use SSML tags for finer control over pronunciation
+            - Max 500 characters per request
+            - Audio will auto-play after generation
+            """)
+
+    # ── Button Actions ───────────────────────────────────────
+    submit_btn.click(
+        fn=run_tts,
+        inputs=[text_input],
+        outputs=[audio_output, status_msg]
+    )
+    clear_btn.click(
+        fn=lambda: ("", None, ""),
+        outputs=[text_input, audio_output, status_msg]
+    )
+
+    gr.Markdown("---")
+
+    # ── Bottom Section: Audio Samples ───────────────────────
+    gr.Markdown("### 📂 Audio Samples")
+    gr.Markdown("Reference outputs from our test set — no SSML tags used.")
+
+    for i, sample in enumerate(SAMPLES):
+        with gr.Row(elem_classes=["sample-row"]):
+            with gr.Column(scale=3):
+                gr.Markdown(f"**Sample {i+1}**")
+                gr.Markdown(f"*\"{sample['transcript']}\"*")
+            with gr.Column(scale=2):
+                gr.Audio(
+                    value=sample["file"],
+                    label="",
+                    interactive=False,
+                    show_download_button=False,
+                )
+
+demo.launch()

requirements.txt

@@ -0,0 +1,2 @@
+gradio>=4.0.0
+requests>=2.28.0