app.py

import gradio as gr

from tts import VachanaFlowTTS
import soundfile as sf

# --- Model initialization
ONNX_DIR = "onnx_model"
DEVICE = "cpu"
tts = VachanaFlowTTS(ONNX_DIR, device=DEVICE)
sample_rate = 44100

def synthesize(text, spk_id=0, step=8, speed=0.9, variation=0.667, normalize=True):
    audio = tts.synthesize(text, spk_id, step, speed, variation, normalize)
    out_wav = "output.wav"
    sf.write(out_wav, audio , sample_rate)
    return out_wav, "✅ สร้างเสียงสำเร็จ!"

custom_css = """

@import url('https://fonts.googleapis.com/css2?family=Sarabun:wght@300;400;600&family=Space+Mono:wght@400;700&display=swap');



:root {

    --bg: #0d0d0f;

    --surface: #16161a;

    --surface2: #1e1e24;

    --border: #2a2a35;

    --accent: #c8f55a;

    --accent2: #5af5c8;

    --text: #f0f0f5;

    --muted: #6b6b80;

    --radius: 14px;

}



body, .gradio-container {

    background: var(--bg) !important;

    font-family: 'Sarabun', sans-serif !important;

    color: var(--text) !important;

}



.gradio-container {

    max-width: 700px !important;

    margin: 0 auto !important;

    padding: 2rem 1.5rem !important;

}

#header {

    text-align: center;

    margin-bottom: 2.5rem;

    padding: 2.5rem 2rem;

    background: var(--surface);

    border: 1px solid var(--border);

    border-radius: 20px;

    position: relative;

    overflow: hidden;

}

#header::before {

    content: '';

    position: absolute;

    top: -60px; right: -60px;

    width: 200px; height: 200px;

    background: radial-gradient(circle, rgba(200,245,90,0.12), transparent 70%);

    border-radius: 50%;

}

#header h1 {

    font-family: 'Space Mono', monospace !important;

    font-size: 2rem !important;

    font-weight: 700 !important;

    color: var(--accent) !important;

    margin: 0 0 0.4rem 0 !important;

    letter-spacing: -1px;

}

#header p {

    color: var(--muted) !important;

    font-size: 0.95rem !important;

    margin: 0 !important;

}

.block, .form {

    background: var(--surface) !important;

    border: 1px solid var(--border) !important;

    border-radius: var(--radius) !important;

}

label span {

    color: var(--muted) !important;

    font-size: 0.8rem !important;

    font-family: 'Space Mono', monospace !important;

    text-transform: uppercase !important;

    letter-spacing: 1px !important;

}

textarea, input[type="text"] {

    background: var(--surface2) !important;

    border: 1px solid var(--border) !important;

    color: var(--text) !important;

    border-radius: 10px !important;

    font-family: 'Sarabun', sans-serif !important;

    font-size: 1rem !important;

    line-height: 1.7 !important;

}

textarea:focus, input:focus {

    border-color: var(--accent) !important;

    box-shadow: 0 0 0 2px rgba(200,245,90,0.12) !important;

}

.radio-group label, .checkbox-group label {

    background: var(--surface2) !important;

    border: 1px solid var(--border) !important;

    border-radius: 8px !important;

    color: var(--text) !important;

    padding: 0.6rem 1.2rem !important;

    cursor: pointer !important;

    transition: all 0.2s ease !important;

}

.radio-group label:hover {

    border-color: var(--accent) !important;

    color: var(--accent) !important;

}

input[type="radio"]:checked + label,

.radio-group label.selected {

    border-color: var(--accent) !important;

    color: var(--accent) !important;

    background: rgba(200,245,90,0.08) !important;

}

button.primary {

    background: var(--accent) !important;

    color: #0d0d0f !important;

    border: none !important;

    border-radius: 10px !important;

    font-family: 'Space Mono', monospace !important;

    font-weight: 700 !important;

    font-size: 0.9rem !important;

    letter-spacing: 1px !important;

    padding: 0.8rem 2rem !important;

    width: 100% !important;

    cursor: pointer !important;

    transition: opacity 0.2s, transform 0.15s !important;

}

button.primary:hover {

    opacity: 0.88 !important;

    transform: translateY(-1px) !important;

}

.upload-area {

    border: 2px dashed var(--border) !important;

    border-radius: 10px !important;

    background: var(--surface2) !important;

    transition: border-color 0.2s !important;

}

.upload-area:hover {

    border-color: var(--accent2) !important;

}

audio {

    width: 100% !important;

    border-radius: 10px !important;

    background: var(--surface2) !important;

}

.status-box {

    font-family: 'Space Mono', monospace !important;

    font-size: 0.82rem !important;

    color: var(--accent2) !important;

    background: var(--surface2) !important;

    border: 1px solid var(--border) !important;

    border-radius: 8px !important;

    padding: 0.6rem 1rem !important;

}

.gap { gap: 1rem !important; }

footer { display: none !important; }

"""

with gr.Blocks(css=custom_css, title="Thai TTS") as demo:
    gr.HTML("""

    <div id="header">

        <h1>Vachana-Flow-TTS</h1>

        <p>Thai Text-to-Speech · Fast with Flow Matching with CPU,GPU</p>

    </div>

    """)
    with gr.Column(elem_classes="gap"):
        text_input = gr.Textbox(
            label="ข้อความ",
            placeholder="พิมพ์ข้อความภาษาไทยที่ต้องการแปลงเป็นเสียง...",
            lines=4,
            value="วันนี้อากาศปลอดโปร่ง ลมพัดเย็นสบาย รู้สึกเหมาะกับการออกไปเดินเล่นหรือจิบกาแฟข้างนอกมากเลย"
        )
        speaker_id = gr.Radio(label="Speaker ID",choices=[0,1],value=0)
        step = gr.Slider(label="N Step",value=8, step=1,minimum=4, maximum=15)
        speed = gr.Slider(label="Speed",value=0.9,step=0.01,minimum=0.3,maximum=1.8)
        variation = gr.Slider(label="Variation",value=0.667,step=0.001, minimum=0.1,maximum=0.8)
        normalize = gr.Checkbox(label="Normalize Audio", value=True)
        run_btn = gr.Button("▶ สร้างเสียง", variant="primary")
    with gr.Column(elem_classes="gap"):
        audio_output = gr.Audio(label="เสียงที่สร้างขึ้น", type="filepath")
        status_output = gr.Textbox(label="สถานะ", elem_classes="status-box", show_label=False, interactive=False)

    run_btn.click(
        fn=synthesize,
        inputs=[text_input, speaker_id, step, speed, variation, normalize],
        outputs=[audio_output, status_output],
    )

demo.launch()