Update app.py

app.py

@@ -1,63 +1,221 @@
-import gradio as gr
+# ============================================================
+# Hugging Face Space - Bilingual TTS with Piper
+# ============================================================
+
+from __future__ import annotations
+
 import os
-import subprocess
+import wave
 import tempfile
+from functools import lru_cache
+from pathlib import Path
+
+import gradio as gr
+from huggingface_hub import hf_hub_download
+from piper import PiperVoice, SynthesisConfig
+
+# ------------------------------------------------------------
+# Config
+# ------------------------------------------------------------
+
+REPO_ID = "rhasspy/piper-voices"
+
+# إضافة المزيد من الأصوات (عربي، بريطاني، أمريكي)
+VOICE_PRESETS = {
+    "Woman (British)": {
+        "label": "Woman (British) — Alba",
+        "model": "en/en_GB/alba/medium/en_GB-alba-medium.onnx",
+        "config": "en/en_GB/alba/medium/en_GB-alba-medium.onnx.json",
+    },
+    "Man (British)": {
+        "label": "Man (British) — Alan",
+        "model": "en/en_GB/alan/medium/en_GB-alan-medium.onnx",
+        "config": "en/en_GB/alan/medium/en_GB-alan-medium.onnx.json",
+    },
+    "Child (British)": {
+        "label": "Child (British, approximate) — Aru",
+        "model": "en/en_GB/aru/medium/en_GB-aru-medium.onnx",
+        "config": "en/en_GB/aru/medium/en_GB-aru-medium.onnx.json",
+    },
+    "Woman (US English)": {
+        "label": "Woman (US English) — Lessac",
+        "model": "en/en_US/lessac/medium/en_US-lessac-medium.onnx",
+        "config": "en/en_US/lessac/medium/en_US-lessac-medium.onnx.json",
+    },
+    "Man (Arabic)": {
+        "label": "Man (Arabic) — Khalil",
+        "model": "ar/ar_JO/khalil/medium/ar_JO-khalil-medium.onnx",
+        "config": "ar/ar_JO/khalil/medium/ar_JO-khalil-medium.onnx.json",
+    }
+}
+
+DEFAULT_TEXT = "Hello. Welcome to our text to speech demo."
+
+CUSTOM_CSS = """
+footer {display:none !important;}
+.gradio-container {
+    max-width: 980px !important;
+}
+.hero {
+    text-align: center;
+    margin-bottom: 10px;
+}
+.note {
+    font-size: 0.95rem;
+    color: #666;
+}
+"""
+
+# ------------------------------------------------------------
+# Voice loading
+# ------------------------------------------------------------
+
+def download_voice_files(preset_name: str) -> tuple[str, str]:
+    preset = VOICE_PRESETS[preset_name]
+    model_path = hf_hub_download(
+        repo_id=REPO_ID,
+        filename=preset["model"],
+        repo_type="model",
+    )
+    config_path = hf_hub_download(
+        repo_id=REPO_ID,
+        filename=preset["config"],
+        repo_type="model",
+    )
+    return model_path, config_path
+
+@lru_cache(maxsize=8)
+def get_voice(preset_name: str) -> PiperVoice:
+    model_path, _ = download_voice_files(preset_name)
+    return PiperVoice.load(model_path)
+
+# ------------------------------------------------------------
+# TTS
+# ------------------------------------------------------------
+
+def synthesize_tts(
+    text: str,
+    preset_name: str,
+    speed: float,
+    volume: float,
+    sentence_silence: float,
+) -> tuple[str | None, str]:
+    text = (text or "").strip()
+    if not text:
+        return None, "Please enter some text."
 
-MODEL_FILE = "en_US-lessac-medium.onnx"
-CONFIG_FILE = "en_US-lessac-medium.onnx.json"
+    try:
+        voice = get_voice(preset_name)
 
-MODEL_URL = "https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/medium/en_US-lessac-medium.onnx"
-CONFIG_URL = "https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/medium/en_US-lessac-medium.onnx.json"
+        safe_speed = max(0.6, min(1.6, float(speed)))
+        length_scale = 1.0 / safe_speed
 
-def download_file(url, filename):
-    if not os.path.exists(filename):
-        result = subprocess.run(
-            ["wget", "-O", filename, url],
-            capture_output=True,
-            text=True
+        syn_config = SynthesisConfig(
+            volume=float(volume),
+            length_scale=length_scale,
+            noise_scale=0.667,
+            noise_w_scale=0.8,
+            sentence_silence=float(sentence_silence),
+            normalize_audio=True,
         )
-        if result.returncode != 0:
-            raise gr.Error("فشل تحميل الملف: " + filename + "\n" + result.stderr)
 
-def generate_audio(text):
-    if not text or not text.strip():
-        raise gr.Error("اكتبي نص أولًا.")
+        tmp = tempfile.NamedTemporaryFile(suffix=".wav", delete=False)
+        tmp_path = tmp.name
+        tmp.close()
 
-    download_file(MODEL_URL, MODEL_FILE)
-    download_file(CONFIG_URL, CONFIG_FILE)
+        with wave.open(tmp_path, "wb") as wav_file:
+            voice.synthesize_wav(
+                text,
+                wav_file,
+                syn_config=syn_config,
+            )
 
-    output_file = tempfile.NamedTemporaryFile(delete=False, suffix=".wav").name
+        return tmp_path, f"Done: {VOICE_PRESETS[preset_name]['label']}"
+    except Exception as error:
+        return None, f"Error: {error}"
+
+# ------------------------------------------------------------
+# Small helpers
+# ------------------------------------------------------------
+
+def fill_example_woman() -> tuple[str, str]:
+    return ("Good morning. This website now speaks with a British voice.", "Woman (British)")
+
+def fill_example_man() -> tuple[str, str]:
+    return ("The dictionary, music, and narration can all use British pronunciation.", "Man (British)")
+
+def fill_example_child() -> tuple[str, str]:
+    return ("Hello. I am the lighter British preset for younger sounding speech.", "Child (British)")
+
+def fill_example_arabic() -> tuple[str, str]:
+    return ("مرحباً بكم، هذا النص مثال على الصوت العربي الجاهز للاستخدام.", "Man (Arabic)")
+
+# ------------------------------------------------------------
+# UI
+# ------------------------------------------------------------
+
+with gr.Blocks(css=CUSTOM_CSS, theme=gr.themes.Soft(), title="Bilingual TTS") as demo:
+    gr.HTML("""
+    <div class="hero">
+        <h1>Bilingual TTS (English & Arabic)</h1>
+        <p class="note">
+            Piper-based Hugging Face Space
+        </p>
+    </div>
+    """)
+
+    with gr.Row():
+        with gr.Column(scale=3):
+            text_input = gr.Textbox(
+                label="Text",
+                value=DEFAULT_TEXT,
+                lines=8,
+                placeholder="Type the text you want spoken...",
+            )
+
+        with gr.Column(scale=2):
+            preset_dropdown = gr.Dropdown(
+                choices=list(VOICE_PRESETS.keys()),
+                value="Woman (British)",
+                label="Voice Preset",
+            )
+
+            speed_slider = gr.Slider(minimum=0.6, maximum=1.6, value=1.0, step=0.05, label="Speed")
+            volume_slider = gr.Slider(minimum=0.2, maximum=1.5, value=1.0, step=0.05, label="Volume")
+            silence_slider = gr.Slider(minimum=0.0, maximum=1.0, value=0.2, step=0.05, label="Sentence Silence")
+
+            generate_btn = gr.Button("Generate Speech", variant="primary")
+
+    with gr.Row():
+        woman_btn = gr.Button("Preset: Woman (EN)")
+        man_btn = gr.Button("Preset: Man (EN)")
+        child_btn = gr.Button("Preset: Child (EN)")
+        arabic_btn = gr.Button("Preset: Man (AR)")
+
+    with gr.Row():
+        audio_output = gr.Audio(
+            label="Generated Audio",
+            type="filepath",
+            format="wav",
+            autoplay=False,
+            interactive=False,
+        )
+
+    status_output = gr.Textbox(
+        label="Status",
+        interactive=False,
+    )
 
-    command = f'echo "{text}" | piper --model {MODEL_FILE} --config {CONFIG_FILE} --output_file {output_file}'
+    woman_btn.click(fn=fill_example_woman, outputs=[text_input, preset_dropdown])
+    man_btn.click(fn=fill_example_man, outputs=[text_input, preset_dropdown])
+    child_btn.click(fn=fill_example_child, outputs=[text_input, preset_dropdown])
+    arabic_btn.click(fn=fill_example_arabic, outputs=[text_input, preset_dropdown])
 
-    result = subprocess.run(
-        command,
-        shell=True,
-        capture_output=True,
-        text=True
+    generate_btn.click(
+        fn=synthesize_tts,
+        inputs=[text_input, preset_dropdown, speed_slider, volume_slider, silence_slider],
+        outputs=[audio_output, status_output],
     )
 
-    if result.returncode != 0:
-        raise gr.Error("Piper فشل في توليد الصوت:\n" + result.stderr)
-
-    if not os.path.exists(output_file) or os.path.getsize(output_file) < 1000:
-        raise gr.Error("تم إنشاء ملف صوت لكنه فارغ أو صغير جدًا.")
-
-    return output_file
-
-demo = gr.Interface(
-    fn=generate_audio,
-    inputs=gr.Textbox(
-        label="Text",
-        value="Hello Thuraya. Welcome to Under the Palm Tree.",
-        lines=3
-    ),
-    outputs=gr.Audio(
-        label="Audio",
-        type="filepath",
-        autoplay=True
-    ),
-    title="Piper TTS Test"
-)
-
-demo.launch()
+if __name__ == "__main__":
+    demo.launch()