tts_gui.py

#!/usr/bin/env python3
"""Silero TTS v5 GUI — синтез русской речи."""

import os
import sys
import threading
import time
import tkinter as tk
from tkinter import ttk

import soundfile as sf

SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
if SCRIPT_DIR not in sys.path:
    sys.path.insert(0, SCRIPT_DIR)
from tts import SileroTTS, SPEAKER_ALIASES  # noqa: E402

# ─── Цвета (в стиле ASR GUI) ──────────────────────────────────────────

BG = '#1a1a2e'
FG = '#e0e0e0'
TEXT_BG = '#16213e'
BTN_BG = '#0f3460'
BTN_STOP = '#e94560'
BTN_CLR = '#555'
ACCENT = '#00d2ff'

# ─── TTS ───────────────────────────────────────────────────────────────

_tts = None


def get_tts():
    global _tts
    if _tts is None:
        _tts = SileroTTS(use_accentor=True)
    return _tts


# ─── Приложение ────────────────────────────────────────────────────────


class TTSApp:
    def __init__(self):
        self.root = tk.Tk()
        self.root.title('Silero TTS v5 — Русский синтез речи')
        self.root.geometry('800x650')
        self.root.configure(bg=BG)

        self._speaker_var = tk.StringVar(value='ru_eduard')
        self._status_var = tk.StringVar(value='Загрузка...')

        self._build_ui()
        self._bind_events()

        # Загрузка модели в фоне
        self.root.after(100, self._load_model)

    def _build_ui(self):
        # ── Верхняя панель: заголовок ──
        header = tk.Label(
            self.root, text='🎙️ Silero TTS v5 — Синтез русской речи',
            font=('Arial', 16, 'bold'), bg=BG, fg=ACCENT,
        )
        header.pack(fill=tk.X, padx=15, pady=(15, 5))

        # ── Текстовое поле ввода ──
        self.text = tk.Text(
            self.root, font=('Arial', 16), bg=TEXT_BG, fg=FG,
            relief=tk.FLAT, border=0, height=6, wrap=tk.WORD,
            insertbackground=FG,
        )
        self.text.pack(fill=tk.BOTH, expand=True, padx=15, pady=10)
        self.text.insert('1.0', 'Введите текст для озвучивания...')
        self.text.bind('<FocusIn>', self._on_focus_in)

        # ── Статус ──
        self.status = tk.Label(
            self.root, textvariable=self._status_var,
            font=('Arial', 12), bg=BG, fg='#aaa',
        )
        self.status.pack(fill=tk.X, padx=15)

        # ── Панель управления ──
        ctrl = tk.Frame(self.root, bg=BG)
        ctrl.pack(fill=tk.X, padx=15, pady=(10, 5))

        # Голос
        tk.Label(ctrl, text='Голос:', font=('Arial', 13),
                 bg=BG, fg=FG).pack(side=tk.LEFT, padx=(0, 5))
        self.speaker_combo = ttk.Combobox(
            ctrl, textvariable=self._speaker_var,
            font=('Arial', 12), state='readonly', width=20,
        )
        self.speaker_combo.pack(side=tk.LEFT, padx=(0, 15))

        # Кнопки
        self.btn_speak = tk.Button(
            ctrl, text='🔊  СИНТЕЗ', font=('Arial', 18, 'bold'),
            bg=BTN_BG, fg='white', relief=tk.FLAT,
            cursor='hand2', height=1, padx=20,
            command=self._click_speak,
        )
        self.btn_speak.pack(side=tk.LEFT, padx=5)

        self.btn_save = tk.Button(
            ctrl, text='💾  СОХРАНИТЬ', font=('Arial', 18, 'bold'),
            bg=BTN_BG, fg='white', relief=tk.FLAT,
            cursor='hand2', height=1, padx=20,
            command=self._click_save,
        )
        self.btn_save.pack(side=tk.LEFT, padx=5)

        self.btn_stop = tk.Button(
            ctrl, text='⏹  СТОП', font=('Arial', 18, 'bold'),
            bg=BTN_STOP, fg='white', relief=tk.FLAT,
            cursor='hand2', state=tk.DISABLED, height=1, padx=20,
            command=self._click_stop,
        )
        self.btn_stop.pack(side=tk.LEFT, padx=5)

        self.btn_clear = tk.Button(
            ctrl, text='Очистить', font=('Arial', 14),
            bg=BTN_CLR, fg='white', relief=tk.FLAT,
            cursor='hand2', height=1,
            command=self._click_clear,
        )
        self.btn_clear.pack(side=tk.LEFT, padx=5)

        # ── Информационная панель внизу ──
        info_frame = tk.Frame(self.root, bg=BG)
        info_frame.pack(fill=tk.X, padx=15, pady=(5, 15))

        self.info_text = tk.Label(
            info_frame,
            text=('Модель: v5_cis_base | 48 kHz | WAV | '
                  'Ударения: ✓ | RTF ~0.04 CPU'),
            font=('Arial', 10), bg=BG, fg='#888',
        )
        self.info_text.pack()

        self._saved_audio = None  # последний синтезированный аудио

    def _bind_events(self):
        self.root.protocol('WM_DELETE_WINDOW', self._on_close)
        self.root.bind('<Control-Return>', lambda e: self._click_speak())
        self.root.bind('<Control-s>', lambda e: self._click_save())

    def _on_focus_in(self, event):
        if self.text.get('1.0', 'end-1c') == 'Введите текст для озвучивания...':
            self.text.delete('1.0', tk.END)

    def _load_model(self):
        def load():
            try:
                tts = get_tts()
                speakers = tts.speakers
                display_names = []
                for s in speakers:
                    aliases = [k for k, v in SPEAKER_ALIASES.items() if v == s]
                    label = s
                    if aliases:
                        label += f'  ({", ".join(aliases)})'
                    display_names.append(label)
                self.root.after(0, lambda: self._on_model_ready(display_names))
            except Exception as e:
                self.root.after(0, lambda: self._status_var.set(
                    f'Ошибка загрузки: {e}'))

        threading.Thread(target=load, daemon=True).start()

    def _on_model_ready(self, display_names):
        self.speaker_combo['values'] = display_names
        self.speaker_combo.current(display_names.index(
            next(n for n in display_names if n.startswith('ru_eduard'))
        ))
        self._status_var.set('Готов. Введите текст и нажмите СИНТЕЗ')
        self.btn_speak.config(state=tk.NORMAL)
        self.btn_save.config(state=tk.NORMAL)

    def _get_speaker(self) -> str:
        selection = self.speaker_combo.get()
        return selection.split()[0]  # берём только ru_имя до скобок

    def _synthesize(self, text: str, speaker: str) -> float:
        """Синтезировать и вернуть длительность аудио."""
        t0 = time.perf_counter()
        self._saved_audio = get_tts().speak(text, speaker=speaker)
        dt = time.perf_counter() - t0
        duration = len(self._saved_audio) / get_tts().sample_rate
        return dt, duration

    def _click_speak(self):
        text = self.text.get('1.0', 'end-1c').strip()
        if not text or text == 'Введите текст для озвучивания...':
            self._status_var.set('Введите текст для синтеза')
            return

        speaker = self._get_speaker()
        self._set_busy(True, f'🔊 Синтезирую: {speaker}...')
        self.root.update()

        def task():
            try:
                dt, duration = self._synthesize(text, speaker)
                # Воспроизвести
                import sounddevice as sd
                sd.play(self._saved_audio, get_tts().sample_rate)
                sd.wait()
                self.root.after(0, lambda: self._on_done(dt, duration, speaker))
            except Exception as e:
                self.root.after(0, lambda: self._on_error(str(e)))

        threading.Thread(target=task, daemon=True).start()

    def _click_save(self):
        text = self.text.get('1.0', 'end-1c').strip()
        if not text or text == 'Введите текст для озвучивания...':
            self._status_var.set('Введите текст для синтеза')
            return

        speaker = self._get_speaker()
        self._set_busy(True, f'💾 Синтезирую: {speaker}...')
        self.root.update()

        def task():
            try:
                dt, duration = self._synthesize(text, speaker)
                # Сохранить
                fname = f'tts_{speaker}_{int(time.time())}.wav'
                sf.write(fname, self._saved_audio, get_tts().sample_rate)
                fsize = os.path.getsize(fname)
                self.root.after(0, lambda: self._on_saved(
                    fname, fsize, dt, duration, speaker))
            except Exception as e:
                self.root.after(0, lambda: self._on_error(str(e)))

        threading.Thread(target=task, daemon=True).start()

    def _click_stop(self):
        import sounddevice as sd
        sd.stop()
        self._set_busy(False, '⏹ Воспроизведение остановлено')

    def _click_clear(self):
        self.text.delete('1.0', tk.END)
        self._status_var.set('Готов')

    def _set_busy(self, busy: bool, msg: str = ''):
        state = tk.DISABLED if busy else tk.NORMAL
        self.btn_speak.config(state=state)
        self.btn_save.config(state=state)
        self.btn_stop.config(state=tk.NORMAL if busy else tk.DISABLED)
        if msg:
            self._status_var.set(msg)
        self.root.update()

    def _on_done(self, dt: float, duration: float, speaker: str):
        rtf = dt / duration if duration > 0 else 0
        self._set_busy(False)
        self._status_var.set(
            f'✅ Воспроизведено | {speaker} | {duration:.2f}s | '
            f'{dt:.3f}s | RTF {rtf:.3f} (×{1/rtf:.1f})'
        )

    def _on_saved(self, fname: str, fsize: int,
                  dt: float, duration: float, speaker: str):
        rtf = dt / duration if duration > 0 else 0
        self._set_busy(False)
        status = (
            f'💾 Сохранено: {fname} ({fsize/1024:.0f} KB) | '
            f'{speaker} | {duration:.2f}s | {dt:.3f}s | RTF {rtf:.3f}'
        )
        self._status_var.set(status)
        # Копируем имя файла в буфер обмена
        self.root.clipboard_clear()
        self.root.clipboard_append(fname)

    def _on_error(self, err: str):
        self._set_busy(False)
        self._status_var.set(f'❌ Ошибка: {err}')

    def _on_close(self):
        import sounddevice as sd
        sd.stop()
        self.root.destroy()

    def run(self):
        self.root.mainloop()


# ─── Main ──────────────────────────────────────────────────────────────

def main():
    app = TTSApp()
    app.run()


if __name__ == '__main__':
    main()