Update app.py

app.py

@@ -0,0 +1,199 @@
+# -*- coding: utf-8 -*-
+
+import os, re, time, random
+import requests
+import gradio as gr
+from pydub import AudioSegment
+import natsort
+
+# ================== PATH SAFE ==================
+BASE_DIR = os.getcwd()
+VOICE_DIR = os.path.join(BASE_DIR, "voices")
+os.makedirs(VOICE_DIR, exist_ok=True)
+
+# ================== CORE ==================
+def check_api_key(api_key):
+    try:
+        r = requests.get(
+            "https://api.elevenlabs.io/v1/user",
+            headers={"xi-api-key": api_key},
+            timeout=10
+        )
+        if r.status_code == 200:
+            sub = r.json().get("subscription", {})
+            return {
+                "valid": True,
+                "remaining": sub.get("character_limit", 0)
+                - sub.get("character_count", 0)
+            }
+        return {"valid": False}
+    except:
+        return {"valid": False}
+
+
+def parse_text_blocks(text, max_len=200):
+    blocks, cur = [], ""
+    for s in re.split(r'(?<=[.!?])\s+', text):
+        if len(cur) + len(s) <= max_len:
+            cur += " " + s
+        else:
+            blocks.append(cur.strip())
+            cur = s
+    if cur:
+        blocks.append(cur.strip())
+    return blocks
+
+
+def estimate_credit(text):
+    return len(text) + 50
+
+
+def generate_voice(text, api_key, voice_id, model_id,
+                   stability, similarity, style, speed, boost):
+
+    time.sleep(random.uniform(1, 2))
+
+    url = f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}"
+    headers = {
+        "xi-api-key": api_key,
+        "Content-Type": "application/json"
+    }
+
+    payload = {
+        "text": text,
+        "model_id": model_id,
+        "voice_settings": {
+            "stability": stability,
+            "similarity_boost": similarity,
+            "style": style,
+            "speed": speed,
+            "use_speaker_boost": boost
+        }
+    }
+
+    r = requests.post(url, headers=headers, json=payload, timeout=30)
+    if r.status_code == 200:
+        return r.content
+    return None
+
+
+def merge_audio(fmt):
+    files = natsort.natsorted([
+        f for f in os.listdir(VOICE_DIR)
+        if f.endswith("." + fmt.lower())
+    ])
+
+    if not files:
+        return None
+
+    combined = AudioSegment.from_file(os.path.join(VOICE_DIR, files[0]))
+    for f in files[1:]:
+        combined += AudioSegment.silent(500)
+        combined += AudioSegment.from_file(os.path.join(VOICE_DIR, f))
+
+    out_path = os.path.join(BASE_DIR, f"output_full.{fmt.lower()}")
+    combined.export(out_path, format=fmt.lower())
+    return out_path
+
+
+# ================== MAIN ==================
+def run_tts(api_keys_text, voice_id, text, model_id, fmt,
+            stability, similarity, style, speed, boost):
+
+    for f in os.listdir(VOICE_DIR):
+        os.remove(os.path.join(VOICE_DIR, f))
+
+    api_keys = [k.strip() for k in api_keys_text.splitlines() if k.strip()]
+    valid_keys = []
+
+    for k in api_keys:
+        info = check_api_key(k)
+        if info.get("valid") and info["remaining"] > 500:
+            valid_keys.append([k, info["remaining"]])
+
+    if not valid_keys:
+        return "❌ Không có API key hợp lệ", None
+
+    blocks = parse_text_blocks(text)
+    key_idx = 0
+
+    for i, block in enumerate(blocks):
+        success = False
+
+        while valid_keys:
+            key, remain = valid_keys[key_idx]
+            need = estimate_credit(block)
+
+            if remain < need:
+                valid_keys.pop(key_idx)
+                continue
+
+            audio = generate_voice(
+                block, key, voice_id, model_id,
+                stability, similarity, style, speed, boost
+            )
+
+            if audio:
+                path = os.path.join(
+                    VOICE_DIR, f"voice_{i+1:03d}.{fmt.lower()}"
+                )
+                with open(path, "wb") as f:
+                    f.write(audio)
+                valid_keys[key_idx][1] -= need
+                success = True
+                break
+            else:
+                valid_keys.pop(key_idx)
+
+        if not success:
+            return "❌ Hết API key khi đang chạy", None
+
+    merged = merge_audio(fmt)
+    return "✅ Hoàn tất", merged
+
+
+# ================== UI ==================
+with gr.Blocks() as demo:
+    gr.Markdown("## 🔊 ElevenLabs TTS – Hugging Face Stable")
+
+    api_keys = gr.Textbox(label="API Keys (mỗi dòng 1 key)", lines=4)
+    voice_id = gr.Textbox(label="Voice ID")
+    text = gr.Textbox(label="Text", lines=6)
+
+    model = gr.Dropdown(
+        choices=[
+            ("Turbo v2.5", "eleven_turbo_v2_5"),
+            ("Flash v2.5", "eleven_flash_v2_5"),
+            ("Multilingual v2", "eleven_multilingual_v2"),
+        ],
+        value="eleven_multilingual_v2",
+        label="Model"
+    )
+
+    fmt = gr.Dropdown(["MP3", "WAV"], value="MP3", label="Format")
+
+    stability = gr.Slider(0, 1, 0.9, label="Stability")
+    similarity = gr.Slider(0, 1, 0.5, label="Similarity")
+    style = gr.Slider(0, 1, 0.4, label="Style")
+    speed = gr.Slider(0.7, 1.2, 0.81, label="Speed")
+    boost = gr.Checkbox(True, label="Speaker Boost")
+
+    run_btn = gr.Button("🎧 Tạo giọng nói")
+    status = gr.Textbox(label="Status")
+    output_audio = gr.Audio(type="filepath", label="Output")
+
+    run_btn.click(
+        run_tts,
+        inputs=[
+            api_keys, voice_id, text, model, fmt,
+            stability, similarity, style, speed, boost
+        ],
+        outputs=[status, output_audio]
+    )
+
+# ⚠️ DÒNG QUAN TRỌNG NHẤT CHO HUGGING FACE
+demo.launch(
+    server_name="0.0.0.0",
+    server_port=7860,
+    show_error=True
+)