app.py

import os
import sys
import importlib.util
import site
import json
import torch
import gradio as gr
import torchaudio
import numpy as np
from huggingface_hub import snapshot_download, hf_hub_download
import subprocess
import re
import spaces
import uuid
import soundfile as sf

# منابع ضروری
downloaded_resources = {
    "configs": False,
    "tokenizer_vq8192": False,
    "fmt_Vq8192ToMels": False,
    "vocoder": False
}

def install_espeak():
    try:
        result = subprocess.run(["which", "espeak-ng"], capture_output=True, text=True)
        if result.returncode != 0:
            print("Installing espeak-ng...")
            subprocess.run(["apt-get", "update"], check=True)
            subprocess.run(["apt-get", "install", "-y", "espeak-ng", "espeak-ng-data"], check=True)
    except Exception as e:
        print(f"Error installing espeak-ng: {e}")

install_espeak()

def patch_langsegment_init():
    try:
        spec = importlib.util.find_spec("LangSegment")
        if spec is None or spec.origin is None: return
        init_path = os.path.join(os.path.dirname(spec.origin), '__init__.py')
        if not os.path.exists(init_path):
            for site_pkg_path in site.getsitepackages():
                potential_path = os.path.join(site_pkg_path, 'LangSegment', '__init__.py')
                if os.path.exists(potential_path):
                    init_path = potential_path
                    break
            else: return

        with open(init_path, 'r') as f: lines = f.readlines()
        modified = False
        new_lines = []
        target_line_prefix = "from .LangSegment import"

        for line in lines:
            if line.strip().startswith(target_line_prefix) and ('setLangfilters' in line or 'getLangfilters' in line):
                mod_line = line.replace(',setLangfilters', '').replace(',getLangfilters', '')
                mod_line = mod_line.replace('setLangfilters,', '').replace('getLangfilters,', '').rstrip(',') 
                new_lines.append(mod_line + '\n')
                modified = True
            else:
                new_lines.append(line)

        if modified:
            with open(init_path, 'w') as f: f.writelines(new_lines)
            try:
                import LangSegment
                importlib.reload(LangSegment)
            except: pass
    except: pass

patch_langsegment_init()

if not os.path.exists("Amphion"):
    subprocess.run(["git", "clone", "https://github.com/open-mmlab/Amphion.git"])
    os.chdir("Amphion")
else:
    if not os.getcwd().endswith("Amphion"):
        os.chdir("Amphion")

if os.path.dirname(os.path.abspath("Amphion")) not in sys.path:
    sys.path.append(os.path.dirname(os.path.abspath("Amphion")))

os.makedirs("wav", exist_ok=True)
os.makedirs("ckpts/Vevo", exist_ok=True)

from models.vc.vevo.vevo_utils import VevoInferencePipeline

def save_audio_pcm16(waveform, output_path, sample_rate=24000):
    try:
        if isinstance(waveform, torch.Tensor):
            waveform = waveform.detach().cpu()
            if waveform.dim() == 2 and waveform.shape[0] == 1:
                waveform = waveform.squeeze(0)
            waveform = waveform.numpy()
        sf.write(output_path, waveform, sample_rate, subtype='PCM_16')
    except Exception as e:
        print(f"Save error: {e}")
        raise e

def setup_configs():
    if downloaded_resources["configs"]: return
    config_path = "models/vc/vevo/config"
    os.makedirs(config_path, exist_ok=True)
    config_files = ["Vq8192ToMels.json", "Vocoder.json"]
    
    for file in config_files:
        file_path = f"{config_path}/{file}"
        if not os.path.exists(file_path):
            try:
                file_data = hf_hub_download(repo_id="amphion/Vevo", filename=f"config/{file}", repo_type="model")
                subprocess.run(["cp", file_data, file_path])
            except Exception as e: print(f"Error downloading config {file}: {e}")
    downloaded_resources["configs"] = True

setup_configs()

device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
print(f"Using device: {device}")

inference_pipelines = {}

def preload_all_resources():
    print("Preloading resources...")
    setup_configs()
    global downloaded_content_style_tokenizer_path, downloaded_fmt_path, downloaded_vocoder_path
    
    if not downloaded_resources["tokenizer_vq8192"]:
        local_dir = snapshot_download(repo_id="amphion/Vevo", repo_type="model", cache_dir="./ckpts/Vevo", allow_patterns=["tokenizer/vq8192/*"])
        downloaded_content_style_tokenizer_path = local_dir
        downloaded_resources["tokenizer_vq8192"] = True
    
    if not downloaded_resources["fmt_Vq8192ToMels"]:
        local_dir = snapshot_download(repo_id="amphion/Vevo", repo_type="model", cache_dir="./ckpts/Vevo", allow_patterns=["acoustic_modeling/Vq8192ToMels/*"])
        downloaded_fmt_path = local_dir
        downloaded_resources["fmt_Vq8192ToMels"] = True
    
    if not downloaded_resources["vocoder"]:
        local_dir = snapshot_download(repo_id="amphion/Vevo", repo_type="model", cache_dir="./ckpts/Vevo", allow_patterns=["acoustic_modeling/Vocoder/*"])
        downloaded_vocoder_path = local_dir
        downloaded_resources["vocoder"] = True
    print("Resources ready.")

downloaded_content_style_tokenizer_path = None
downloaded_fmt_path = None
downloaded_vocoder_path = None

preload_all_resources()

def get_pipeline():
    if "timbre" in inference_pipelines:
        return inference_pipelines["timbre"]
    
    pipeline = VevoInferencePipeline(
        content_style_tokenizer_ckpt_path=os.path.join(downloaded_content_style_tokenizer_path, "tokenizer/vq8192"),
        fmt_cfg_path="./models/vc/vevo/config/Vq8192ToMels.json",
        fmt_ckpt_path=os.path.join(downloaded_fmt_path, "acoustic_modeling/Vq8192ToMels"),
        vocoder_cfg_path="./models/vc/vevo/config/Vocoder.json",
        vocoder_ckpt_path=os.path.join(downloaded_vocoder_path, "acoustic_modeling/Vocoder"),
        device=device,
    )
    
    inference_pipelines["timbre"] = pipeline
    return pipeline

@spaces.GPU()
def vevo_timbre(content_wav, reference_wav):
    session_id = str(uuid.uuid4())[:8]
    temp_content_path = f"wav/c_{session_id}.wav"
    temp_reference_path = f"wav/r_{session_id}.wav"
    output_path = f"wav/out_{session_id}.wav"
    
    if content_wav is None or reference_wav is None:
        raise ValueError("Please upload audio files")
    
    try:
        # --- آماده سازی Reference (اول رفرنس را پردازش می‌کنیم تا سطح صدا را بگیریم) ---
        if isinstance(reference_wav, tuple):
            ref_sr, ref_data = reference_wav if isinstance(reference_wav[0], int) else (reference_wav[1], reference_wav[0])
        else:
            ref_sr, ref_data = reference_wav

        if len(ref_data.shape) > 1 and ref_data.shape[1] > 1:
            ref_data = np.mean(ref_data, axis=1)

        ref_tensor = torch.FloatTensor(ref_data).unsqueeze(0)
        if ref_sr != 24000:
            ref_tensor = torchaudio.functional.resample(ref_tensor, ref_sr, 24000)
            ref_sr = 24000
        
        # محاسبه انرژی رفرنس
        ref_max_vol = torch.max(torch.abs(ref_tensor)) + 1e-6
        ref_tensor = ref_tensor / ref_max_vol * 0.95 # نرمال سازی رفرنس
        
        # برش رفرنس به 20 ثانیه
        if ref_tensor.shape[1] > 24000 * 20:
             ref_tensor = ref_tensor[:, :24000 * 20]

        save_audio_pcm16(ref_tensor, temp_reference_path, ref_sr)

        # --- آماده سازی Content ---
        if isinstance(content_wav, tuple):
            content_sr, content_data = content_wav if isinstance(content_wav[0], int) else (content_wav[1], content_wav[0])
        else:
            content_sr, content_data = content_wav

        if len(content_data.shape) > 1 and content_data.shape[1] > 1:
            content_data = np.mean(content_data, axis=1)
        
        content_tensor = torch.FloatTensor(content_data).unsqueeze(0)
        if content_sr != 24000:
            content_tensor = torchaudio.functional.resample(content_tensor, content_sr, 24000)
            content_sr = 24000
        
        # نرمال سازی هوشمند: صدای ورودی را هم‌سطح صدای رفرنس می‌کنیم
        content_tensor = content_tensor / (torch.max(torch.abs(content_tensor)) + 1e-6) * 0.95

        # --- منطق Chunking ---
        pipeline = get_pipeline()
        
        SR = 24000
        CHUNK_LEN = 10 * SR
        OVERLAP = 1 * SR 
        INPUT_SIZE = CHUNK_LEN + OVERLAP 
        
        total_samples = content_tensor.shape[1]
        print(f"[{session_id}] High Quality Processing (64 Steps)... Duration: {total_samples/SR:.2f}s")
        
        final_parts = []
        overlap_buffer = None
        
        for start in range(0, total_samples, CHUNK_LEN):
            end_input = min(start + INPUT_SIZE, total_samples)
            
            current_input_chunk = content_tensor[:, start:end_input]
            save_audio_pcm16(current_input_chunk, temp_content_path, SR)
            
            try:
                gen = pipeline.inference_fm(
                    src_wav_path=temp_content_path,
                    timbre_ref_wav_path=temp_reference_path,
                    flow_matching_steps=64,  # <--- کیفیت بالا (قبلاً 32 بود)
                )
                
                if torch.isnan(gen).any(): gen = torch.nan_to_num(gen, nan=0.0)
                if gen.dim() == 1: gen = gen.unsqueeze(0)
                gen = gen.cpu().squeeze(0).numpy()
                
                current_len = len(gen)
                
                if overlap_buffer is not None:
                    mix_len = len(overlap_buffer)
                    if current_len < mix_len:
                        mix_len = current_len
                        overlap_buffer = overlap_buffer[:mix_len]
                    
                    head_to_mix = gen[:mix_len]
                    body_rest = gen[mix_len:]
                    
                    alpha = np.linspace(0, 1, mix_len)
                    blended_segment = (overlap_buffer * (1 - alpha)) + (head_to_mix * alpha)
                    
                    final_parts.append(blended_segment)
                    
                    if len(body_rest) > OVERLAP:
                        pure_body = body_rest[:-OVERLAP]
                        final_parts.append(pure_body)
                        overlap_buffer = body_rest[-OVERLAP:]
                    else:
                        final_parts.append(body_rest)
                        overlap_buffer = None
                
                else:
                    if current_len > OVERLAP:
                        final_parts.append(gen[:-OVERLAP])
                        overlap_buffer = gen[-OVERLAP:]
                    else:
                        final_parts.append(gen)
                        overlap_buffer = None

            except Exception as e:
                print(f"Error in chunk: {e}")
                missing_len = end_input - start
                if overlap_buffer is not None:
                    missing_len -= len(overlap_buffer)
                    final_parts.append(overlap_buffer)
                    overlap_buffer = None
                final_parts.append(np.zeros(max(0, missing_len)))

        if overlap_buffer is not None:
            final_parts.append(overlap_buffer)
            
        if len(final_parts) > 0:
            full_audio = np.concatenate(final_parts)
        else:
            full_audio = np.zeros(24000)

        save_audio_pcm16(full_audio, output_path, SR)
        return output_path

    finally:
        if os.path.exists(temp_content_path): os.remove(temp_content_path)
        if os.path.exists(temp_reference_path): os.remove(temp_reference_path)

with gr.Blocks(title="Vevo-Timbre (Ultra Quality)") as demo:
    gr.Markdown("## Vevo-Timbre: Zero-Shot Voice Conversion (Ultra Quality)")
    gr.Markdown("""
    **ویژگی‌ها:**
    - **Steps 64:** کیفیت و دقت بافت صدا دو برابر شده است.
    - **Auto-Leveling:** سطح صدای شما با مدل تنظیم می‌شود.
    - **Seamless Stitching:** بدون پرش و بدون اضافه شدن زمان.
    
    **نکته مهم:** برای بهترین نتیجه، سعی کنید **لحن، سرعت و احساس** صدای خودتان را شبیه فایل هدف کنید. مدل فقط جنس صدا را تغییر می‌دهد، نه بازیگری شما را!
    """)
    
    with gr.Row():
        with gr.Column():
            timbre_content = gr.Audio(label="Source Audio (صدای شما)", type="numpy")
            timbre_reference = gr.Audio(label="Target Timbre (صدای هدف)", type="numpy")
            timbre_button = gr.Button("Generate (Ultra Quality)", variant="primary")
        with gr.Column():
            timbre_output = gr.Audio(label="Result")
            
    timbre_button.click(vevo_timbre, inputs=[timbre_content, timbre_reference], outputs=timbre_output)

demo.launch()