update

.gitignore

@@ -1,2 +1,111 @@
 flagged/
 .vscode/
+.vscode
+.ipynb_checkpoints
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+.mypy
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# celery beat schedule file
+celerybeat-schedule
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+
+.vscode

app.py

@@ -1,81 +1,309 @@
-from logging import getLogger
+import shutil
+import sys
+import zipfile
+from datetime import datetime
+from functools import partial
+from pathlib import Path
+from tempfile import gettempdir
+from typing import Dict
 
 import gradio as gr
-import numpy as np
-import numpy.typing as npt
-from en_tts import Synthesizer, Transcriber
+from scipy.io.wavfile import read
 
-FLOAT32_64_MIN_WAV = -1.0
-FLOAT32_64_MAX_WAV = 1.0
-INT16_MIN = np.iinfo(np.int16).min  # -32768 = -(2**15)
-INT16_MAX = np.iinfo(np.int16).max  # 32767 = 2**15 - 1
-INT32_MIN = np.iinfo(np.int32).min  # -2147483648 = -(2**31)
-INT32_MAX = np.iinfo(np.int32).max  # 2147483647 = 2**31 - 1
+from en_tts_app import (get_log_path, get_work_dir, initialize_app, load_models_to_cache, run_main,
+                        synthesize_english)
 
 
-def synt(text: str) -> str:
-    logger = getLogger(__name__)
-    logger.info("Initializing transcriber...")
-    transcriber = Transcriber()
+def run():
+  exit_code = initialize_app()
+  if exit_code > 0:
+    sys.exit(exit_code)
 
-    logger.info("Initializing synthesizer...")
-    synthesizer = Synthesizer()
+  exit_code = run_main(launch_interface)
+  sys.exit(exit_code)
 
-    logger.info("Transcribing...")
-    text_ipa = transcriber.transcribe_to_ipa(text)
 
-    logger.info("Synthesizing...")
-    audio = synthesizer.synthesize(text_ipa)
-    audio_int = convert_wav(audio, np.int16)
-    return 22050, audio_int
+def launch_interface():
+  cache = load_models_to_cache()
 
+  fn = partial(synt, cache=cache)
 
-def get_max_value(dtype):
-    # see wavfile.write() max positive eg. on 16-bit PCM is 32767
-    if dtype == np.int16:
-        return INT16_MAX
+  # iface = gr.Interface(
+  #   fn=fn,
+  #   inputs=[gr.Textbox(example_text, label="Text")],
+  #   outputs=[gr.Audio(type="numpy", label="Speech", autoplay=True)],
+  # )
 
-    if dtype == np.int32:
-        return INT32_MAX
+  with gr.Blocks(
+    title="en-tts"
+  ) as demo:
+    gr.Markdown(
+      """
+      # English Speech Synthesis
 
-    if dtype in (np.float32, np.float64):
-        return FLOAT32_64_MAX_WAV
+      Enter or paste your text into the provided text box and click the **Synthesize** button to convert it into speech. You can adjust settings as desired before synthesizing.
+      """
+    )
 
-    assert False
+    with gr.Tab("Synthesis"):
+      with gr.Row():
+        with gr.Column():
+          with gr.Group():
+            input_txt_box = gr.Textbox(
+              None,
+              label="Input",
+              placeholder="Enter the text you want to synthesize (or load an example from below).",
+              lines=10,
+              max_lines=5000,
+            )
 
+            with gr.Accordion("Settings", open=False):
+              sent_norm_check_box = gr.Checkbox(
+                False,
+                label="Skip normalization",
+                info="Skip normalization of numbers, units and abbreviations."
+              )
 
-def get_min_value(dtype):
-    if dtype == np.int16:
-        return INT16_MIN
+              sent_sep_check_box = gr.Checkbox(
+                False,
+                label="Skip sentence separation",
+                info="Skip sentence separation after these characters: .?!"
+              )
 
-    if dtype == np.int32:
-        return INT32_MIN
+              sil_sent_txt_box = gr.Number(
+                0.4,
+                minimum=0.0,
+                maximum=60,
+                step=0.1,
+                label="Silence between sentences (s)",
+                info="Insert silence between each sentence."
+              )
 
-    if dtype in (np.float32, np.float64):
-        return FLOAT32_64_MIN_WAV
+              sil_para_txt_box = gr.Number(
+                1.0,
+                minimum=0.0,
+                maximum=60,
+                step=0.1,
+                label="Silence between paragraphs (s)",
+                info="Insert silence between each paragraph."
+              )
 
-    assert False
+              seed_txt_box = gr.Number(
+                0,
+                minimum=0,
+                maximum=999999,
+                label="Seed",
+                info="Seed used for inference in order to be able to reproduce the results."
+              )
 
+              sigma_txt_box = gr.Number(
+                1.0,
+                minimum=0.0,
+                maximum=1.0,
+                step=0.001,
+                label="Sigma",
+                info="Sigma used for inference in WaveGlow."
+              )
 
-def convert_wav(wav: npt.NDArray[np.float64], to_dtype):
-    '''
-    if the wav is over-amplified the result will also be over-amplified.
-    '''
-    if wav.dtype != to_dtype:
-        wav = wav / (-1 * get_min_value(wav.dtype)) * get_max_value(to_dtype)
-        if to_dtype in (np.int16, np.int32):
-            # the default seems to be np.fix instead of np.round on wav.astype()
-            wav = np.round(wav, 0)
-        wav = wav.astype(to_dtype)
+              max_decoder_steps_txt_box = gr.Number(
+                5000,
+                minimum=1,
+                step=500,
+                label="Maximum decoder steps",
+                info="Stop the synthesis after this number of decoder steps at the latest."
+              )
 
-    return wav
+              denoiser_txt_box = gr.Number(
+                0.005,
+                minimum=0.0,
+                maximum=1.0,
+                step=0.001,
+                label="Denoiser strength",
+                info="Level of noise reduction used to remove the noise bias from WaveGlow."
+              )
 
+          synt_btn = gr.Button("Synthesize", variant="primary")
 
-example_text = "When the sunlight strikes raindrops in the air, they act as a prism and form a rainbow."
+        with gr.Column():
+          with gr.Group():
 
-iface = gr.Interface(
-    fn=synt,
-    inputs=[gr.Textbox(example_text, label="Text")],
-    outputs=[gr.Audio(type="numpy", label="Speech", autoplay=True)],
-)
-iface.launch()
+            with gr.Row():
+              with gr.Column():
+                out_audio = gr.Audio(
+                  type="numpy",
+                  label="Output",
+                  autoplay=True,
+                )
+
+                with gr.Accordion(
+                    "Log",
+                    open=False,
+                  ):
+                  out_md = gr.Textbox(
+                    interactive=False,
+                    show_copy_button=True,
+                    lines=15,
+                    max_lines=10000,
+                    placeholder="Log will be displayed here.",
+                    show_label=False,
+                  )
+
+                  dl_btn = gr.DownloadButton(
+                    "Download working directory",
+                    variant="secondary",
+                  )
+
+      with gr.Row():
+        gr.Examples(
+          examples=[
+            [
+              "When the sunlight strikes raindrops in the air, they act as a prism and form a rainbow.",
+              5000, 1.0, 0.0005, 0, 0.4, 1.0, False, False
+            ],
+            [
+              "Please call Stella. Ask her to bring these things with her from the store: six spoons of fresh snow peas, five thick slabs of blue cheese, and maybe a snack for her brother Bob.\n\nWe also need a small plastic snake and a big toy frog for the kids. She can scoop these things into three red bags, and we will go meet her Wednesday at the train station.",
+              5000, 1.0, 0.0005, 0, 0.4, 1.0, False, False
+            ],
+            # [
+            #   "The North Wind and the Sun were disputing which was the stronger, when a traveler came along wrapped in a warm cloak. They agreed that the one who first succeeded in making the traveler take his cloak off should be considered stronger than the other. Then the North Wind blew as hard as he could, but the more he blew the more closely did the traveler fold his cloak around him; and at last the North Wind gave up the attempt. Then the Sun shined out warmly, and immediately the traveler took off his cloak. And so the North Wind was obliged to confess that the Sun was the stronger of the two.",
+            # ],
+            # [
+            #   "When the sunlight strikes raindrops in the air, they act as a prism and form a rainbow. The rainbow is a division of white light into many beautiful colors. These take the shape of a long round arch, with its path high above, and its two ends apparently beyond the horizon. There is, according to legend, a boiling pot of gold at one end. People look, but no one ever finds it. When a man looks for something beyond his reach, his friends say he is looking for the pot of gold at the end of the rainbow. Throughout the centuries people have explained the rainbow in various ways. Some have accepted it as a miracle without physical explanation. To the Hebrews it was a token that there would be no more universal floods. The Greeks used to imagine that it was a sign from the gods to foretell war or heavy rain. The Norsemen considered the rainbow as a bridge over which the gods passed from earth to their home in the sky. Others have tried to explain the phenomenon physically. Aristotle thought that the rainbow was caused by reflection of the sun's rays by the rain. Since then physicists have found that it is not reflection, but refraction by the raindrops which causes the rainbows. Many complicated ideas about the rainbow have been formed. The difference in the rainbow depends considerably upon the size of the drops, and the width of the colored band increases as the size of the  rops increases. The actual primary rainbow observed is said to be the effect of superimposition of a number of bows. If the red of the second bow falls upon the green of the first, the result is to give a bow with an abnormally wide yellow band, since red and green light when mixed form yellow. This is a very common type of bow, one showing mainly red and yellow, with little or no green or blue."
+            # ]
+          ],
+          fn=fn,
+          inputs=[
+            input_txt_box,
+            max_decoder_steps_txt_box,
+            sigma_txt_box,
+            denoiser_txt_box,
+            seed_txt_box,
+            sil_sent_txt_box,
+            sil_para_txt_box,
+            sent_norm_check_box,
+            sent_sep_check_box,
+          ],
+          outputs=[
+            out_audio,
+            out_md,
+            dl_btn,
+          ],
+          label="Examples",
+          cache_examples=False,
+        )
+
+    with gr.Tab("Info"):
+      with gr.Column():
+        gr.Markdown(
+          """
+          ### General information
+
+          - Speaker: Linda Johnson
+          - Language: English
+          - Accent: North American
+          - Supported special characters: `.?!,:;-—"'()[]`
+
+          ### Evaluation results
+
+          |Metric|Value|
+          |---|---|
+          |MOS naturalness|3.55 ± 0.28 (GT: 4.17 ± 0.23)|
+          |MOS intelligibility|4.44 ± 0.24 (GT: 4.63 ± 0.19)|
+          |Mean MCD-DTW|29.15|
+          |Mean penalty|0.1018|
+
+          ### Components
+
+          |Component|Name|URLs|
+          |---|---|---|
+          |Acoustic model|Tacotron|[Checkpoint](https://zenodo.org/records/10107104), [Code](https://github.com/stefantaubert/tacotron)|
+          |Vocoder|WaveGlow|[Checkpoint](https://catalog.ngc.nvidia.com/orgs/nvidia/models/waveglow_ljs_256channels/files?version=3), [Code](https://github.com/stefantaubert/waveglow)
+          |Dataset|LJ Speech|[Link](https://keithito.com/LJ-Speech-Dataset), [Transcriptions](https://zenodo.org/records/7499098)|
+
+          ### Citation
+
+          Taubert, S. (2024). en-tts (Version 0.0.1) [Computer software]. https://doi.org/10.5281/zenodo.10479347
+
+          ### Acknowledgments
+
+          Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 416228727 – [CRC 1410](https://gepris.dfg.de/gepris/projekt/416228727?context=projekt&task=showDetail&id=416228727)
+
+          The authors gratefully acknowledge the GWK support for funding this project by providing computing time through the Center for Information Services and HPC (ZIH) at TU Dresden.
+
+          The authors are grateful to the Center for Information Services and High Performance Computing [Zentrum fur Informationsdienste und Hochleistungsrechnen (ZIH)] at TU Dresden for providing its facilities for high throughput calculations.
+
+          ### App information
+
+          - Version: 0.0.1
+          - License: [MIT](https://github.com/stefantaubert/en-tts?tab=MIT-1-ov-file#readme)
+          - GitHub: [stefantaubert/en-tts](https://github.com/stefantaubert/en-tts)
+          """
+        )
+
+    synt_btn.click(
+      fn=fn,
+      inputs=[
+        input_txt_box,
+        max_decoder_steps_txt_box,
+        sigma_txt_box,
+        denoiser_txt_box,
+        seed_txt_box,
+        sil_sent_txt_box,
+        sil_para_txt_box,
+        sent_norm_check_box,
+        sent_sep_check_box,
+      ],
+      outputs=[
+        out_audio,
+        out_md,
+        dl_btn,
+      ],
+    )
+
+  demo.launch(
+    share=False,
+    debug=True,
+    inbrowser=True,
+    quiet=False,
+    show_api=False,
+  )
+
+
+def synt(text: str, max_decoder_steps: int, sigma: float, denoiser_strength: float, seed: int, silence_sentences: float, silence_paragraphs: float, skip_normalization: bool, skip_sentence_separation: bool, cache: Dict) -> str:
+  result_path = synthesize_english(
+    text, cache,
+    max_decoder_steps=max_decoder_steps,
+    seed=seed,
+    sigma=sigma,
+    denoiser_strength=denoiser_strength,
+    silence_paragraphs=silence_paragraphs,
+    silence_sentences=silence_sentences,
+    skip_normalization=skip_normalization,
+    skip_sentence_separation=skip_sentence_separation,
+  )
+
+  rate, audio_int = read(result_path)
+  logs = get_log_path().read_text("utf-8")
+  zip_dl_path = create_zip_file_of_output()
+  return (rate, audio_int), logs, zip_dl_path
+
+
+def create_zip_file_of_output() -> Path:
+  work_dir = get_work_dir()
+
+  name = f"en-tts-{datetime.now().strftime('%Y-%m-%dT%H-%M-%S')}"
+
+  res = shutil.make_archive(Path(gettempdir()) / name, 'zip', root_dir=work_dir)
+
+  resulting_zip = Path(res)
+
+  with zipfile.ZipFile(resulting_zip, "a", compression=zipfile.ZIP_DEFLATED) as zipf:
+    source_path = get_log_path()
+    destination = 'output.log'
+    zipf.write(source_path, destination)
+
+  return resulting_zip
+
+
+if __name__ == "__main__":
+  run()

app_old.py

@@ -0,0 +1,81 @@
+from logging import getLogger
+
+import gradio as gr
+import numpy as np
+import numpy.typing as npt
+from en_tts import Synthesizer, Transcriber
+
+FLOAT32_64_MIN_WAV = -1.0
+FLOAT32_64_MAX_WAV = 1.0
+INT16_MIN = np.iinfo(np.int16).min  # -32768 = -(2**15)
+INT16_MAX = np.iinfo(np.int16).max  # 32767 = 2**15 - 1
+INT32_MIN = np.iinfo(np.int32).min  # -2147483648 = -(2**31)
+INT32_MAX = np.iinfo(np.int32).max  # 2147483647 = 2**31 - 1
+
+logger = getLogger(__name__)
+logger.info("Initializing transcriber...")
+transcriber = Transcriber()
+
+logger.info("Initializing synthesizer...")
+synthesizer = Synthesizer()
+
+
+def synt(text: str) -> str:
+    logger.info("Transcribing...")
+    text_ipa = transcriber.transcribe_to_ipa(text)
+
+    logger.info("Synthesizing...")
+    audio = synthesizer.synthesize(text_ipa)
+    audio_int = convert_wav(audio, np.int16)
+    return 22050, audio_int
+
+
+def get_max_value(dtype):
+    # see wavfile.write() max positive eg. on 16-bit PCM is 32767
+    if dtype == np.int16:
+        return INT16_MAX
+
+    if dtype == np.int32:
+        return INT32_MAX
+
+    if dtype in (np.float32, np.float64):
+        return FLOAT32_64_MAX_WAV
+
+    assert False
+
+
+def get_min_value(dtype):
+    if dtype == np.int16:
+        return INT16_MIN
+
+    if dtype == np.int32:
+        return INT32_MIN
+
+    if dtype in (np.float32, np.float64):
+        return FLOAT32_64_MIN_WAV
+
+    assert False
+
+
+def convert_wav(wav: npt.NDArray[np.float64], to_dtype):
+    '''
+    if the wav is over-amplified the result will also be over-amplified.
+    '''
+    if wav.dtype != to_dtype:
+        wav = wav / (-1 * get_min_value(wav.dtype)) * get_max_value(to_dtype)
+        if to_dtype in (np.int16, np.int32):
+            # the default seems to be np.fix instead of np.round on wav.astype()
+            wav = np.round(wav, 0)
+        wav = wav.astype(to_dtype)
+
+    return wav
+
+
+example_text = "When the sunlight strikes raindrops in the air, they act as a prism and form a rainbow."
+
+iface = gr.Interface(
+    fn=synt,
+    inputs=[gr.Textbox(example_text, label="Text")],
+    outputs=[gr.Audio(type="numpy", label="Speech", autoplay=True)],
+)
+iface.launch()

en_tts_app/__init__.py

@@ -0,0 +1,3 @@
+from en_tts_app.app import initialize_app, run_main
+from en_tts_app.globals import get_conf_dir, get_log_path, get_work_dir, reset_log
+from en_tts_app.main import load_models_to_cache, synthesize_english

en_tts_app/app.py

@@ -0,0 +1,90 @@
+import logging
+import sys
+from logging import getLogger
+from time import perf_counter
+from typing import Callable
+
+from en_tts_app.globals import get_conf_dir, get_log_path
+from en_tts_app.logging_configuration import (configure_app_logger, configure_file_logger,
+                                              configure_root_logger, get_file_logger)
+from en_tts_app.main import load_models_to_cache
+
+INITIALIZED = False
+
+
+def ensure_conf_dir_exists():
+  conf_dir = get_conf_dir()
+  if not conf_dir.is_dir():
+    root_logger = getLogger()
+    root_logger.debug("Creating configuration directory ...")
+    conf_dir.mkdir(parents=False, exist_ok=False)
+
+
+def configure_external_loggers() -> None:
+  file_logger = get_file_logger()
+  for logger_name in ("httpcore", "httpx", "asyncio", "matplotlib"):
+    logger = getLogger(logger_name)
+    logger.parent = file_logger
+    logger.disabled = False
+    logger.propagate = True
+    logger.level = logging.DEBUG
+
+
+def initialize_app() -> int:
+  global INITIALIZED
+  assert not INITIALIZED
+  # CLI logging = INFO
+  # External loggers go to file-logger
+  # file-logger = DEBUG
+
+  # # disable mpl temporarily
+  # mpl_logger = getLogger("matplotlib")
+  # mpl_logger.disabled = True
+  # mpl_logger.propagate = False
+
+  configure_root_logger(logging.INFO)
+  root_logger = getLogger()
+
+  logfile = get_log_path()
+  try:
+    configure_file_logger(logfile, logging.DEBUG)
+  except Exception as ex:
+    root_logger.exception("Logging to file is not possible. Exiting.", exc_info=ex, stack_info=True)
+    return 1
+
+  configure_app_logger(logging.INFO)
+  configure_external_loggers()
+
+  ensure_conf_dir_exists()
+
+  # path not encapsulated in "" because it is only console out
+  root_logger.info(f"Log will be written to: {logfile.absolute()}")
+  INITIALIZED = True
+  return 0
+
+
+def run_main(method: Callable) -> int:
+  global INITIALIZED
+  assert INITIALIZED
+
+  flogger = get_file_logger()
+
+  start = perf_counter()
+  success = True
+
+  try:
+    method()
+  except ValueError as error:
+    success = False
+    logger = getLogger(__name__)
+    logger.debug("ValueError occurred.", exc_info=error)
+  except Exception as error:
+    success = False
+    logger = getLogger(__name__)
+    logger.debug("Exception occurred.", exc_info=error)
+
+  duration = perf_counter() - start
+  flogger.debug(f"Total duration (seconds): {duration}")
+
+  exit_code = 0 if success else 1
+  return exit_code

en_tts_app/globals.py

@@ -0,0 +1,34 @@
+import shutil
+from pathlib import Path
+from tempfile import gettempdir
+
+from en_tts_cli.logging_configuration import get_cli_logger
+
+
+def get_conf_dir() -> Path:
+  conf_dir = Path.home() / ".en-tts"
+  return conf_dir
+
+
+def get_work_dir() -> Path:
+  work_dir = Path(gettempdir()) / "en-tts"
+  return work_dir
+
+
+def get_log_path() -> Path:
+  return Path(gettempdir()) / "en-tts.log"
+
+
+def reset_log() -> None:
+  get_log_path().write_text("", "utf-8")
+
+
+def reset_work_dir():
+  root_logger = get_cli_logger()
+  work_dir = get_work_dir()
+
+  if work_dir.is_dir():
+    root_logger.debug("Deleting working directory ...")
+    shutil.rmtree(work_dir)
+  root_logger.debug("Creating working directory ...")
+  work_dir.mkdir(parents=False, exist_ok=False)

en_tts_app/helper.py

@@ -0,0 +1,11 @@
+
+import os
+from pathlib import Path
+from typing import Generator
+
+
+def get_all_files_in_all_subfolders(directory: Path) -> Generator[Path, None, None]:
+  for root, _, files in os.walk(directory):
+    for name in files:
+      file_path = Path(root) / name
+      yield file_path

en_tts_app/logging_configuration.py

@@ -0,0 +1,143 @@
+import logging
+import os
+import platform
+import sys
+from importlib.metadata import version
+from logging import Formatter, Handler, Logger, StreamHandler, getLogger
+from logging.handlers import MemoryHandler
+from pathlib import Path
+from pkgutil import iter_modules
+
+__APP_NAME = "en-tts"
+
+__version__ = version(__APP_NAME)
+
+
+class ConsoleFormatter(logging.Formatter):
+  """Logging colored formatter, adapted from https://stackoverflow.com/a/56944256/3638629"""
+
+  purple = '\x1b[34m'
+  blue = '\x1b[36m'
+  # blue = '\x1b[38;5;39m'
+  yellow = '\x1b[38;5;226m'
+  red = '\x1b[1;49;31m'
+  bold_red = '\x1b[1;49;31m'
+  reset = '\x1b[0m'
+
+  collected_loggers = set()
+
+  def __init__(self):
+    super().__init__()
+    self.datefmt = '%H:%M:%S'
+    fmt = '(%(levelname)s) %(message)s'
+    fmt_info = '%(message)s'
+
+    self.fmts = {
+        logging.NOTSET: self.purple + fmt + self.reset,
+        logging.DEBUG: self.blue + fmt + self.reset,
+        logging.INFO: fmt_info,
+        logging.WARNING: self.yellow + fmt + self.reset,
+        logging.ERROR: self.red + fmt + self.reset,
+        logging.CRITICAL: self.bold_red + fmt + self.reset,
+    }
+
+  def format(self, record):
+    log_fmt = self.fmts.get(record.levelno)
+    formatter = logging.Formatter(log_fmt, self.datefmt)
+    self.collected_loggers.add(record.name)
+
+    return formatter.format(record)
+
+
+def add_console_out(logger: Logger) -> StreamHandler:
+  console = StreamHandler()
+  logger.addHandler(console)
+  set_console_formatter(console)
+  return console
+
+
+def set_console_formatter(handler: Handler) -> None:
+  logging_formatter = ConsoleFormatter()
+  handler.setFormatter(logging_formatter)
+
+
+def set_logfile_formatter(handler: Handler) -> None:
+  fmt = '[%(asctime)s.%(msecs)03d] %(name)s (%(levelname)s) %(message)s'
+  datefmt = '%Y/%m/%d %H:%M:%S'
+  logging_formatter = Formatter(fmt, datefmt)
+  handler.setFormatter(logging_formatter)
+
+
+def get_app_logger() -> Logger:
+  logger = getLogger("en_tts_app")
+  return logger
+
+
+def get_file_logger() -> Logger:
+  flogger = getLogger("file.en_tts_app")
+  return flogger
+
+
+def configure_app_logger(level: int) -> None:
+  app_logger = get_app_logger()
+  app_logger.handlers.clear()
+  assert len(app_logger.handlers) == 0
+  console_handler = add_console_out(app_logger)
+  # console_handler.setLevel(logging.DEBUG if debug else logging.INFO)
+  app_logger.setLevel(level)
+
+  core_logger = getLogger("en_tts")
+  core_logger.parent = app_logger
+
+  file_logger = get_file_logger()
+  app_logger.parent = file_logger
+
+
+def configure_root_logger(level: int) -> None:
+  # productive = False
+  # loglevel = logging.INFO if productive else logging.DEBUG
+  root_logger = getLogger()
+  root_logger.setLevel(level)
+  root_logger.manager.disable = logging.NOTSET
+  if len(root_logger.handlers) > 0:
+    console = root_logger.handlers[0]
+    set_console_formatter(console)
+  else:
+    console = add_console_out(root_logger)
+
+  # console.setLevel(logging.DEBUG if debug else logging.INFO)
+
+
+def configure_file_logger(path: Path, level: int):
+  flogger = get_file_logger()
+  assert len(flogger.handlers) == 0
+  path.parent.mkdir(parents=True, exist_ok=True)
+  if path.is_file():
+    os.remove(path)
+  path.write_text("")
+  fh = logging.FileHandler(path)
+  set_logfile_formatter(fh)
+  # fh.setLevel(logging.DEBUG if debug else logging.INFO)
+  flogger.setLevel(level)
+  # mh = MemoryHandler(buffer_capacity, logging.ERROR, fh, True)
+  flogger.addHandler(fh)
+
+  if flogger.propagate:
+    flogger.propagate = False
+
+
+def log_sysinfo():
+  flogger = get_file_logger()
+
+  sys_version = sys.version.replace('\n', '')
+  flogger.debug(f"CLI version: {__version__}")
+  flogger.debug(f"Python version: {sys_version}")
+  flogger.debug("Modules: %s", ', '.join(sorted(p.name for p in iter_modules())))
+
+  my_system = platform.uname()
+  flogger.debug(f"System: {my_system.system}")
+  flogger.debug(f"Node Name: {my_system.node}")
+  flogger.debug(f"Release: {my_system.release}")
+  flogger.debug(f"Version: {my_system.version}")
+  flogger.debug(f"Machine: {my_system.machine}")
+  flogger.debug(f"Processor: {my_system.processor}")

en_tts_app/main.py

@@ -0,0 +1,157 @@
+import logging
+import shutil
+from pathlib import Path
+from typing import Dict, Optional
+
+from ordered_set import OrderedSet
+from pronunciation_dictionary import PronunciationDict, SerializationOptions, save_dict
+
+from en_tts.helper import get_default_device, normalize_audio
+from en_tts.io import save_audio
+from en_tts.synthesizer import Synthesizer
+from en_tts.transcriber import Transcriber
+from en_tts_app.globals import get_conf_dir, get_work_dir, reset_log, reset_work_dir
+from en_tts_app.logging_configuration import get_app_logger, get_file_logger, log_sysinfo
+
+CACHE_TRANSCRIBER = "transcriber"
+CACHE_SYNTHESIZER = "synthesizer"
+
+
+def load_models_to_cache() -> Dict:
+  cli_logger = get_app_logger()
+  cache = {}
+
+  conf_dir = get_conf_dir()
+  device = get_default_device()
+
+  cli_logger.info("Initializing Transcriber...")
+  cache[CACHE_TRANSCRIBER] = Transcriber(conf_dir)
+
+  cli_logger.info("Initializing Synthesizer...")
+  cache[CACHE_SYNTHESIZER] = Synthesizer(conf_dir, device)
+  return cache
+
+
+def synthesize_english(text: str, cache: Dict, *, max_decoder_steps: int = 5000, sigma: float = 1.0, denoiser_strength: float = 0.0005, seed: int = 0, silence_sentences: float = 0.4, silence_paragraphs: float = 1.0, loglevel: int = 2, skip_normalization: bool = False, skip_sentence_separation: bool = False, custom_output: Optional[Path] = None) -> Path:
+  cli_logger = get_app_logger()
+  reset_log()
+  reset_work_dir()
+  log_sysinfo()
+
+  if loglevel == 0:
+    cli_logger.setLevel(logging.WARNING)
+
+  if custom_output is None:
+    custom_output = get_work_dir() / "output.wav"
+
+  text_ipa = convert_eng_to_ipa(
+    text, cache[CACHE_TRANSCRIBER],
+    loglevel=loglevel,
+    skip_normalization=skip_normalization,
+    skip_sentence_separation=skip_sentence_separation
+  )
+
+  output_path = synthesize_ipa_core(
+    text_ipa, cache[CACHE_SYNTHESIZER], custom_output,
+    max_decoder_steps=max_decoder_steps, sigma=sigma, denoiser_strength=denoiser_strength,
+    seed=seed, silence_sentences=silence_sentences, silence_paragraphs=silence_paragraphs, loglevel=loglevel
+  )
+
+  return output_path
+
+
+def convert_eng_to_ipa(text: str, transcriber: Transcriber, *, loglevel: int = 1, skip_normalization: bool = False, skip_sentence_separation: bool = False) -> str:
+  t = transcriber
+  text_ipa = t.transcribe_to_ipa(text, skip_normalization, skip_sentence_separation)
+
+  if loglevel >= 1:
+    for txt, name in (
+      (text, "text"),
+      (t.text_normed, "text.normed"),
+      (t.text_sentenced, "text.sentenced"),
+      (t.text_ipa, "text.ipa"),
+      (t.text_ipa_readable, "text.ipa.readable"),
+      (text_ipa, "text.ipa.silenced"),
+    ):
+      try_log_text(txt, name)
+
+    for v, name in (
+      (t.vocabulary, "vocabulary"),
+      (t.oov1, "oov1"),
+      (t.oov2, "oov2"),
+      (t.oov3, "oov3"),
+    ):
+      try_log_voc(v, name)
+
+    for d, name in (
+      (t.dict1, "dict1"),
+      (t.dict1_single, "dict1.single"),
+      (t.dict2, "dict2"),
+      (t.dict2_single, "dict2.single"),
+      (t.dict1_2, "dict1+2"),
+      (t.dict3, "dict3"),
+      (t.dict3_single, "dict3.single"),
+      (t.dict1_2_3, "dict1+2+3"),
+      (t.dict4_arpa, "dict4.arpa"),
+      (t.dict4, "dict4"),
+      (t.dict4_single, "dict4.single"),
+      (t.dict1_2_3_4, "dict1+2+3+4"),
+    ):
+      try_log_dict(d, name)
+  return text_ipa
+
+
+def synthesize_ipa_core(text_ipa: str, synthesizer: Synthesizer, output: Path, *, max_decoder_steps: int = 5000, sigma: float = 1.0, denoiser_strength: float = 0.0005, seed: int = 0, silence_sentences: float = 0.4, silence_paragraphs: float = 1.0, loglevel: int = 1) -> Path:
+  logger = logging.getLogger(__name__)
+  work_dir = get_work_dir()
+
+  audio = synthesizer.synthesize(text_ipa, max_decoder_steps, seed, sigma,
+                                 denoiser_strength, silence_sentences, silence_paragraphs, silent=loglevel == 0)
+  unnormed_out = work_dir / "output.unnormed.wav"
+  save_audio(audio, unnormed_out)
+  work_dir_output = unnormed_out
+
+  try:
+    normalize_audio(unnormed_out, work_dir / "output.wav")
+    work_dir_output = work_dir / "output.wav"
+  except Exception as error:
+    logger.warning("Normalization was not possible!", exc_info=error, stack_info=True)
+    logger.info(f"Saved audio to: '{unnormed_out.absolute()}'")
+
+  if output != work_dir_output:
+    try:
+      output.parent.mkdir(parents=True, exist_ok=True)
+      shutil.copyfile(work_dir_output, output)
+    except Exception as ex:
+      logger.exception(
+        f"Output couldn't be created at: '{output.absolute()}'", exc_info=ex, stack_info=True)
+      logger.info(f"Saved audio to: '{work_dir_output.absolute()}'")
+  logger.info(f"Saved audio to: '{output.absolute()}'")
+  return output
+
+
+def try_log_dict(dictionary: Optional[PronunciationDict], name: str) -> None:
+  if dictionary:
+    work_dir = get_work_dir()
+    logfile = work_dir / f"{name}.dict"
+    save_dict(dictionary, logfile, "utf-8", SerializationOptions("TAB", False, True))
+    flogger = get_file_logger()
+    flogger.info(f"{name}: {logfile.absolute()}")
+
+
+def try_log_voc(vocabulary: Optional[OrderedSet[str]], name: str) -> None:
+  if vocabulary:
+    work_dir = get_work_dir()
+    logfile = work_dir / f"{name}.txt"
+    logfile.write_text("\n".join(vocabulary), "utf-8")
+    flogger = get_file_logger()
+    flogger.info(f"{name}: {logfile.absolute()}")
+
+
+def try_log_text(text: Optional[str], name: str) -> None:
+  if text:
+    work_dir = get_work_dir()
+    logfile = work_dir / f"{name}.txt"
+    logfile.write_text(text, "utf-8")
+    flogger = get_file_logger()
+    flogger.info(f"{name}: {logfile.absolute()}")