deployment/src/f5_tts/f5_tts_webui.py

"""

F5-TTS Thai WebUI - Refactored Version

เวอร์ชันที่ปรับปรุงโครงสร้างใหม่ให้มีระเบียบและง่ายต่อการดูแลรักษา

"""

import argparse
import sys
import os
import gradio as gr

# Add the src directory to Python path for imports
current_dir = os.path.dirname(os.path.abspath(__file__))
src_dir = os.path.dirname(current_dir)
if src_dir not in sys.path:
    sys.path.insert(0, src_dir)

from f5_tts.model_manager import ModelManager
from f5_tts.tts_processor import TTSProcessor, SpeechToTextProcessor
from f5_tts.multi_speech_processor import MultiSpeechProcessor
from f5_tts.ui_components import UIComponents
from f5_tts.config import MAX_SPEECH_TYPES


class F5TTSWebUI:
    """หลัก Web UI Application สำหรับ F5-TTS Thai"""
    
    def __init__(self):
        self.model_manager = ModelManager()
        self.tts_processor = TTSProcessor(self.model_manager)
        self.stt_processor = SpeechToTextProcessor()
        self.multi_speech_processor = MultiSpeechProcessor(self.model_manager)
        self.ui_components = UIComponents()
        
    def create_gradio_interface(self):
        """สร้าง Gradio interface"""
        with gr.Blocks(title="F5-TTS ไทย", theme=gr.themes.Ocean()) as demo:
            gr.Markdown("# F5-TTS ภาษาไทย")
            gr.Markdown("สร้างคำพูดจากข้อความ ด้วย Zero-shot TTS หรือ เสียงต้นฉบับ ภาษาไทย.")

            # Model selection section
            model_select, model_custom, model_status, load_custom_btn = self.ui_components.create_model_selection_section()

            # Setup model selection events
            self._setup_model_selection_events(
                model_select, model_custom, model_status, load_custom_btn
            )

            # Create tabs
            #with gr.Tab(label="Text To Speech"):
            #    self._create_tts_tab()
            
            with gr.Tab(label="Multi Speech"):
                self._create_multispeech_tab()
            
            #with gr.Tab(label="Speech to Text"):
            #    self._create_stt_tab()

        return demo
    
    def _setup_model_selection_events(self, model_select, model_custom, model_status, load_custom_btn):
        """ตั้งค่า events สำหรับการเลือกโมเดล"""
        
        # Model selection change event
        model_select.change(
            fn=self.model_manager.update_custom_model_visibility,
            inputs=model_select,
            outputs=model_custom
        )
        
        # Load custom model button
        load_custom_btn.click(
            fn=self.model_manager.load_model_by_choice,
            inputs=[model_select, model_custom],
            outputs=model_status
        )
    
    def _create_tts_tab(self):
        """สร้าง Text To Speech tab"""
        tts_components = self.ui_components.create_tts_tab(self.tts_processor.infer_tts)
        
        # Setup TTS generation
        tts_components['controls']['generate_btn'].click(
            fn=self.tts_processor.infer_tts,
            inputs=[
                tts_components['inputs']['ref_audio'],
                tts_components['inputs']['ref_text'],
                tts_components['inputs']['gen_text'],
                tts_components['inputs']['remove_silence'],
                tts_components['inputs']['cross_fade_duration'],
                tts_components['inputs']['nfe_step'],
                tts_components['inputs']['speed'],
                tts_components['inputs']['cfg_strength'],
                tts_components['inputs']['max_chars'],
                tts_components['inputs']['seed'],
                tts_components['inputs']['no_ref_audio']
            ],
            outputs=[
                tts_components['outputs']['output_audio'],
                tts_components['outputs']['spectrogram'],
                tts_components['inputs']['ref_text'],
                tts_components['outputs']['seed_output']
            ]
        )
    
    def _create_multispeech_tab(self):
        """สร้าง Multi Speech tab"""
        ms_components = self.ui_components.create_multispeech_tab()
        
        # Setup speech type management
        self._setup_speech_type_events(ms_components)
        
        # Setup multispeech generation
        self._setup_multispeech_generation(ms_components)
        
        # Setup segment editing
        self._setup_segment_editing(ms_components)
    
    def _setup_speech_type_events(self, ms_components):
        """ตั้งค่า events สำหรับ speech type management"""
        
        # Add speech type button
        ms_components['controls']['add_speech_type_btn'].click(
            fn=self.ui_components.add_speech_type_fn,
            outputs=ms_components['controls']['speech_type_rows']
        )
        
        # Delete speech type buttons
        for i in range(1, len(self.ui_components.speech_type_delete_btns)):
            if self.ui_components.speech_type_delete_btns[i] is not None:
                self.ui_components.speech_type_delete_btns[i].click(
                    fn=self.ui_components.delete_speech_type_fn,
                    outputs=[
                        self.ui_components.speech_type_rows[i], 
                        self.ui_components.speech_type_names[i], 
                        self.ui_components.speech_type_audios[i], 
                        self.ui_components.speech_type_ref_texts[i]
                    ]
                )
        
        # Insert speech type buttons
        for i, insert_btn in enumerate(self.ui_components.speech_type_insert_btns):
            insert_fn = self.ui_components.make_insert_speech_type_fn(i)
            insert_btn.click(
                fn=insert_fn,
                inputs=[ms_components['inputs']['gen_text'], self.ui_components.speech_type_names[i]],
                outputs=ms_components['inputs']['gen_text']
            )
        
        # Validation for generate button
        ms_components['inputs']['gen_text'].change(
            fn=self.multi_speech_processor.validate_speech_types,
            inputs=[ms_components['inputs']['gen_text']] + ms_components['inputs']['speech_type_names'],
            outputs=ms_components['controls']['generate_btn']
        )
    
    def _setup_multispeech_generation(self, ms_components):
        """ตั้งค่า multispeech generation"""
        
        # Prepare inputs for generation
        generation_inputs = [
            ms_components['inputs']['gen_text'],
            ms_components['inputs']['cross_fade_duration'],
            ms_components['inputs']['nfe_step']
        ] + (
            ms_components['inputs']['speech_type_names'] +
            ms_components['inputs']['speech_type_audios'] +
            ms_components['inputs']['speech_type_ref_texts'] +
            [ms_components['inputs']['remove_silence']] +
            ms_components['inputs']['segment_silence_inputs']
        )
        
        # Prepare outputs for generation
        generation_outputs = [
            ms_components['outputs']['audio_output'],
            ms_components['outputs']['download_btn']
        ] + (
            ms_components['outputs']['segment_players'] +
            ms_components['outputs']['segment_text_inputs'] +
            ms_components['outputs']['segment_silence_inputs'] +
            ms_components['outputs']['segment_regen_btns'] +
            [ms_components['state']['segments_state'], ms_components['state']['sr_state']]
        )
        
        # Generate button click
        ms_components['controls']['generate_btn'].click(
            fn=self._wrap_multispeech_generation,
            inputs=generation_inputs,
            outputs=generation_outputs
        )
    
    def _wrap_multispeech_generation(self, gen_text, cross_fade_duration, nfe_step, *args):
        """Wrapper สำหรับ multispeech generation"""
        speech_types_data = args[:MAX_SPEECH_TYPES * 3]
        remove_silence = args[MAX_SPEECH_TYPES * 3]
        silence_inputs = args[MAX_SPEECH_TYPES * 3 + 1:]
        
        return self.multi_speech_processor.generate_multistyle_speech(
            gen_text,
            cross_fade_duration,
            nfe_step,
            speech_types_data,
            remove_silence,
            silence_inputs
        )
    
    def _setup_segment_editing(self, ms_components):
        """ตั้งค่า segment editing"""
        
        # Update silence button
        ms_components['controls']['update_silence_btn'].click(
            fn=self.multi_speech_processor.update_silence_all,
            inputs=ms_components['inputs']['segment_silence_inputs'] + [
                ms_components['state']['segments_state'], 
                ms_components['state']['sr_state']
            ],
            outputs=ms_components['outputs']['segment_players'] + 
                   ms_components['outputs']['segment_text_inputs'] + 
                   ms_components['outputs']['segment_silence_inputs'] + 
                   ms_components['outputs']['segment_regen_btns'] + [
                       ms_components['outputs']['audio_output'],
                       ms_components['outputs']['download_btn'],
                       ms_components['state']['segments_state'],
                       ms_components['state']['sr_state']
                   ]
        )
        
        # Regenerate segment buttons
        for i, btn in enumerate(ms_components['outputs']['segment_regen_btns']):
            btn.click(
                fn=self._wrap_regenerate_segment,
                inputs=[
                    gr.State(i),
                    ms_components['outputs']['segment_text_inputs'][i],
                    ms_components['outputs']['segment_silence_inputs'][i],
                    ms_components['state']['segments_state'],
                    ms_components['inputs']['cross_fade_duration'],
                    ms_components['inputs']['nfe_step']
                ],
                outputs=ms_components['outputs']['segment_players'] + 
                       ms_components['outputs']['segment_text_inputs'] + 
                       ms_components['outputs']['segment_silence_inputs'] + 
                       ms_components['outputs']['segment_regen_btns'] + [
                           ms_components['outputs']['audio_output'],
                           ms_components['outputs']['download_btn'],
                           ms_components['state']['segments_state'],
                           ms_components['state']['sr_state']
                       ]
            )
    
    def _wrap_regenerate_segment(self, idx, new_text, silence_ms, segments, cross_fade_duration, nfe_step):
        """Wrapper สำหรับ regenerate segment"""
        return self.multi_speech_processor.regenerate_segment(
            idx, new_text, silence_ms, segments, cross_fade_duration, nfe_step
        )
    
    def _create_stt_tab(self):
        """สร้าง Speech to Text tab"""
        stt_components = self.ui_components.create_stt_tab()
        
        # Setup STT generation
        stt_components['controls']['generate_btn_stt'].click(
            fn=self.stt_processor.transcribe_text,
            inputs=[
                stt_components['inputs']['ref_audio_input'],
                stt_components['inputs']['is_translate'],
                stt_components['inputs']['model_wp'],
                stt_components['inputs']['compute_type'],
                stt_components['inputs']['target_lg'],
                stt_components['inputs']['source_lg']
            ],
            outputs=stt_components['outputs']['output_ref_text']
        )


def main():
    """Main function สำหรับรัน application"""
    try:
        parser = argparse.ArgumentParser(description="F5-TTS Thai WebUI - Refactored")
        parser.add_argument("--share", action="store_true", help="Share the app")
        args = parser.parse_args()

        print("กำลังเริ่มต้น F5-TTS Thai WebUI...")
        app = F5TTSWebUI()
        demo = app.create_gradio_interface()
        print("WebUI พร้อมใช้งาน!")
        demo.launch(inbrowser=True, share=args.share)
    except Exception as e:
        print(f"เกิดข้อผิดพลาด: {e}")
        import traceback
        traceback.print_exc()


if __name__ == "__main__":
    main()