Refactor

appnew.py

@@ -1,334 +0,0 @@
-from ipywidgets import widgets
-import copy
-import json
-import glob
-import os
-import time
-import threading
-from IPython.display import Audio, display, HTML, FileLink
-from pathlib import Path
-import subprocess
-import shutil
-from google.colab import files
-from ipywidgets import TwoByTwoLayout
-import re
-
-SEGMENT_LENGTH = 60
-
-CUSTOM_MODELS_FILENAME = "customModels"
-CUSTOM_MODELS_DIR = f"/drive/MyDrive/{CUSTOM_MODELS_FILENAME}"
-
-MUSIC_EXTENSIONS = ['.mp3', '.wav', '.flac', '.aac', '.ogg']
-
-SEGMENTS_DIRNAME = f"/content/segments"
-INFERENCE_OUTPUT_DIRNAME = "/content/inference_output"
-
-def progress(value, max=100):
-    return HTML("""
-        <progress
-            value='{value}'
-            max='{max}',
-            style='width: 100%'
-        >
-            {value}
-        </progress>
-    """.format(value=value, max=max))
-
-def is_valid_filename(filename):
-    if re.search(r'[<>:"/\\|?*\x00-\x1f]', filename):
-        return False
-    if re.search(r'[-\s()]', filename):
-        return False
-    return True
-
-def clean_filename(filename):
-    basename = os.path.basename(filename)
-    cleaned_basename = re.sub(r'[%<>:"/\\|?*\x00-\x1f]', '', basename)
-    cleaned_basename =  re.sub(r'[-\s]+', '_', cleaned_basename)
-    cleaned_basename = re.sub(r'[\(\)]+', '', cleaned_basename)
-    cleaned_basename = cleaned_basename.replace("'", "").replace('"', '').replace("$", "")
-    cleaned_basename_arr = cleaned_basename.split()
-    for i in range(len(cleaned_basename_arr)):
-      if i != len(cleaned_basename_arr)-1:
-          cleaned_basename_arr[i] = cleaned_basename_arr[i].replace(".", "")
-    cleaned_basename = " ".join(cleaned_basename_arr)
-    return os.path.join(os.path.dirname(filename), cleaned_basename)
-
-def get_audio_files():
-  audio_files = []
-  for root, dirs, files in os.walk("/content"):
-    for filename in files:
-      file_extension = os.path.splitext(filename)[1]
-      if file_extension.lower() in MUSIC_EXTENSIONS and "output" not in filename:
-        audio_files.append(filename)
-
-  return audio_files
-
-def get_speakers():
-  speakers = []
-
-  for _,dirs,_ in os.walk(CUSTOM_MODELS_DIR):
-    for folder in dirs:
-      cur_speaker = {}
-      # Look for G_****.pth
-      g = glob.glob(os.path.join(CUSTOM_MODELS_DIR,folder,'G_*.pth'))
-      if not len(g):
-        continue
-      cur_speaker["model_path"] = g[0]
-      cur_speaker["model_folder"] = folder
-
-      # Look for *.pt (clustering model)
-      clst = glob.glob(os.path.join(CUSTOM_MODELS_DIR,folder,'*.pt'))
-      if not len(clst):
-        cur_speaker["cluster_path"] = ""
-      else:
-        cur_speaker["cluster_path"] = clst[0]
-
-      # Look for config.json
-      cfg = glob.glob(os.path.join(CUSTOM_MODELS_DIR,folder,'*.json'))
-      if not len(cfg):
-        continue
-      cur_speaker["cfg_path"] = cfg[0]
-      with open(cur_speaker["cfg_path"]) as f:
-        try:
-          cfg_json = json.loads(f.read())
-        except Exception as e:
-          print("Malformed config json in "+folder)
-        for name, i in cfg_json["spk"].items():
-          cur_speaker["name"] = name
-          cur_speaker["id"] = i
-          if not name.startswith('.'):
-            speakers.append(copy.copy(cur_speaker))
-
-  return sorted(speakers, key=lambda x:x["name"].lower())
-
-def slice_audio(filepath):
-    assert os.path.exists(filepath), f"Не удалось найти {filepath}. Убедитесь, что вы ввели правильное имя файла."
-    # Get the filename and extension of the input file
-    filename, extension = os.path.splitext(filepath)
-    filename = filename.split("/")[-1]
-
-    os.makedirs(SEGMENTS_DIRNAME, exist_ok=True)
-
-    # Set the output filename pattern
-    output_pattern = f"{SEGMENTS_DIRNAME}/{filename}_%d{extension}"
-
-    # Use ffmpeg to split the audio into segments
-    os.system(f"ffmpeg -i {filepath} -f segment -segment_time {SEGMENT_LENGTH} -c copy {output_pattern}")
-
-
-def get_container_format(filename):
-    command = ["ffprobe", "-v", "error", "-select_streams", "v:0", "-show_entries", "format=format_name", "-of", "default=noprint_wrappers=1:nokey=1", filename]
-    process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-    output, error = process.communicate()
-    if error:
-        raise ValueError(f"Ошибка при получении формата контейнера: {error.decode()}")
-    return output.decode().strip()
-
-def run_inference(speaker, f0_method, transpose, noise_scale, cluster_ratio, is_pitch_prediction_enabled):
-    loading_bar = display(progress(0, 100), display_id=True)
-    model_path = speaker["model_path"]
-    config_path = speaker["cfg_path"]
-    cluster_path = speaker["cluster_path"]
-
-    all_segs_paths = sorted(Path(SEGMENTS_DIRNAME).glob("*"))
-
-    for index, seg_path in enumerate(all_segs_paths):
-      max_load_value = float((index + 1)/len(all_segs_paths)) * 100
-      loading_bar.update(progress(max_load_value / 2, 100))
-      inference_cmd = f"svc infer {seg_path.absolute()} -m {model_path} -c {config_path} {f'-k {cluster_path} -r {cluster_ratio}' if cluster_path != '' and cluster_ratio > 0 else ''} -t {transpose} --f0-method {f0_method} -n {noise_scale} -o {INFERENCE_OUTPUT_DIRNAME}/{seg_path.name} {'' if is_pitch_prediction_enabled else '--no-auto-predict-f0'}"
-      # print(f"\nPerforming inference on... {seg_path.absolute()}\ninference cmd: {inference_cmd}")
-      result = subprocess.run(
-            inference_cmd.split(),
-            stdout=subprocess.PIPE,
-            stderr=subprocess.STDOUT,
-            text=True
-          )
-      loading_bar.update(progress(max_load_value, 100))
-
-      if result.stderr:
-        if "AttributeError" in result.stderr:
-          raise Exception(result.stderr + "Убедитесь, что ваша модель не 4.0-v2. Этот блокнот работает только на моделях 4.0-v1.")
-
-      files_length = len(sorted(Path(SEGMENTS_DIRNAME).glob("*")))
-      if files_length == 0:
-        raise Exception("Произошла неизвестная ошибка!")
-
-
-def concatenate_segments(final_filename):
-    foldername = Path(INFERENCE_OUTPUT_DIRNAME)
-    assert foldername.exists(), "папка не существует. Введите правильное имя папки"
-    all_segs = [f for f in sorted(foldername.glob("**/*")) if f.is_file()]
-    print(all_segs)
-    try:
-      ext = all_segs[0].suffix
-      with open(foldername/"concat_list.txt", "w") as f:
-        for seg in all_segs:
-            f.write('file ' + str(seg.absolute()) + "\n")
-      os.system(f"ffmpeg -f concat -safe 0 -i {foldername}/concat_list.txt -codec copy {foldername}/{final_filename}")
-    except:
-      raise Exception(f'В каталоге {foldername} не найдено ни одного файла')
-
-def cleanup_dirs():
-  !rm -R {INFERENCE_OUTPUT_DIRNAME} &> /dev/null
-  !rm -R {SEGMENTS_DIRNAME} &> /dev/null
-  !rm -R ./so_vits_svc_fork.log &> /dev/null
-
-class InferenceGui():
-  def __init__(self):
-    # Initialize the background watcher thread as None
-    speakers = get_speakers()
-    self.is_inferencing = False
-    self.final_filename = ""
-    self.speakers = speakers if speakers is not None else []
-    self.speaker_list = [x["name"] for x in self.speakers]
-    self.speaker_dropdown = widgets.Dropdown(
-        options = self.speaker_list,
-        description="AI модель"
-    )
-
-    self.audio_files = get_audio_files()
-    self.audio_files_dropdown = widgets.Dropdown(
-      options = self.audio_files,
-      description="Аудиофайл"
-    )
-
-    self.cluster_ratio_tx = widgets.FloatSlider(
-      value=1,
-      min=0,
-      max=1.0,
-      step=0.05,
-      description='Соотношение кластеров',
-      disabled=False,
-      continuous_update=False,
-      orientation='horizontal',
-      readout=True,
-    )
-
-    self.noise_scale_tx = widgets.FloatSlider(
-      value=2,
-      min=-2,
-      max=2,
-      step=.4,
-      description='Шкала шума',
-      disabled=False,
-      continuous_update=False,
-      orientation='horizontal',
-      readout=True,
-    )
-
-    def convert_cb(btn):
-      if (self.is_inferencing):
-        return
-      self.convert()
-    self.convert_btn = widgets.Button(description="Конвертировать")
-    self.convert_btn.on_click(convert_cb)
-
-    def refresh_files(btn):
-      self.update_file_list_dropdown()
-    self.refresh_files_btn = widgets.Button(description="Обновить аудиофайлы")
-    self.refresh_files_btn.on_click(refresh_files)
-
-    cluster_container = widgets.HBox([self.cluster_ratio_tx, widgets.Label(value="Отрегулируйте соотношение между звучанием, похожим на тембр цели, и четкостью и артикулированностью, чтобы найти подходящий компромисс.")])
-    noise_scale_container = widgets.HBox([self.noise_scale_tx, widgets.Label(value="Если выходной сигнал звучит гулко/металлически, попробуйте увеличить масштаб шума. Если появляются артефакты, похожие на плохое шумоподавление или погружение динамика в воду, уменьшите масштаб шума.")])
-
-    audio_files_container = widgets.HBox([
-        self.audio_files_dropdown,
-        self.refresh_files_btn
-    ])
-
-    voice_cloning_tab = widgets.VBox([self.speaker_dropdown, audio_files_container, cluster_container, noise_scale_container])
-
-    buttons_container = widgets.HBox([self.convert_btn])
-
-    if (len(self.audio_files) == 0):
-      audio_file_error_widget = widgets.HBox([
-         widgets.Label(value='Пожалуйста, загрузите аудиофайл и нажмите кнопку воспроизведения, чтобы повторно запустить эту ячейку.')
-      ])
-      display(audio_file_error_widget)
-      return
-
-    display(voice_cloning_tab)
-    display(buttons_container)
-
-  def update_file_list_dropdown(self):
-        self.audio_files = get_audio_files()
-        self.audio_files_dropdown.options = self.audio_files
-
-  def clean(self):
-    input_filepaths = [f for f in glob.glob('/content/**/*.*', recursive=True)
-    if any(f.endswith(ex) for ex in ['.wav','.flac','.mp3','.ogg','.opus'])]
-    for f in input_filepaths:
-      os.remove(f)
-    subprocess.run(['sudo', 'updatedb'])
-    self.update_file_list_dropdown()
-
-
-  def convert(self):
-    ts0 = time.time()
-
-    # Prevent a conversion process from one starting if one is already running
-    self.is_inferencing = True
-
-    speaker = next(x for x in self.speakers if x["name"] ==
-          self.speaker_dropdown.value)
-    model_path = os.path.join(os.getcwd(),speaker["model_path"])
-    config_path = os.path.join(os.getcwd(),speaker["cfg_path"])
-    cluster_path = os.path.join(os.getcwd(),speaker["cluster_path"])
-    file_path = os.path.join(os.getcwd(), str(self.audio_files_dropdown.value))
-    f0_method = "dio"
-    transpose = 0
-    noise_scale = int(self.noise_scale_tx.value)
-    cluster_ratio = float(self.cluster_ratio_tx.value)
-    is_pitch_prediction_enabled = True
-
-    if not speaker:
-      print("Пожалуйста, выберите модель искусственного интеллекта.")
-      return
-    if not self.audio_files_dropdown.value or self.audio_files_dropdown.value == "":
-      print("Пожалуйста, выберите аудиофайл для клонирования.")
-      return
-
-    if not is_valid_filename(file_path):
-      try:
-        new_filename = clean_filename(file_path)
-        os.rename(file_path, new_filename)
-        file_path = new_filename
-      except:
-        print("Пожалуйста, повторно запустите эту ячейку, нажав кнопку воспроизведения. Произошла неизвестная ошибка.")
-
-    if os.path.exists(SEGMENTS_DIRNAME) or os.path.exists(INFERENCE_OUTPUT_DIRNAME):
-      print(f"Обнаружены предыдущие папки {SEGMENTS_DIRNAME} и {INFERENCE_OUTPUT_DIRNAME}.")
-      cleanup_dirs()
-
-    # SLICE AUDIO
-    slice_audio(file_path)
-
-    # PERFORM INFERENCE
-    os.makedirs("inference_output", exist_ok=True)
-    run_inference(speaker, f0_method, transpose, noise_scale, cluster_ratio, is_pitch_prediction_enabled)
-
-    cleaned_speaker_name = speaker['name'].replace(" ", "_")
-    final_filename = f"{Path(file_path).stem}_{cleaned_speaker_name}_output{Path(file_path).suffix}"
-    self.final_filename = final_filename
-
-    # CONCATENATE FILES IN INFERENCE OUTPUT DIR
-    concatenate_segments(final_filename)
-
-    # MOVE FINAL CONCATENATED FILE TO TOP-LEVEL IN CURRENT DIR
-    shutil.move(Path(INFERENCE_OUTPUT_DIRNAME, final_filename), Path(final_filename))
-
-    # CLEAN UP
-    cleanup_dirs()
-
-    ts1 = time.time()
-    print(f"Total Time Elapsed: {ts1 - ts0} seconds")
-    print(f"\nГотово! Можете скачать выходной файл через проводник как '{final_filename}' или через аудио-плеер ниже.")
-
-    audio = Audio(final_filename, autoplay=False)
-    display(audio)
-    self.is_inferencing = False
-    self.update_file_list_dropdown()
-
-
-gui = InferenceGui()