ttsforge 0.1.1__tar.gz → 0.1.2__tar.gz

{ttsforge-0.1.1 → ttsforge-0.1.2}/.github/workflows/codecov.yml

@@ -12,7 +12,7 @@ jobs:
     - name: Install espeak-ng
       run: |
         sudo apt-get update
-        sudo apt-get install -y espeak-ng
+        sudo apt-get install -y espeak-ng ffmpeg
     - name: 'generate report'
       run: |
         pip install coverage click pytest pytest-cov

{ttsforge-0.1.1 → ttsforge-0.1.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ttsforge
-Version: 0.1.1
+Version: 0.1.2
 Summary: Generate audiobooks from EPUB files using Kokoro ONNX TTS.
 Author-email: Holger Nahrstaedt <nahrstaedt@gmail.com>
 License: MIT License
@@ -617,6 +617,7 @@ ttsforge convert book.epub --gpu
 | `pause_paragraph`           | `0.9`          | Paragraph pause (seconds)            |
 | `pause_variance`            | `0.05`         | Pause variance (seconds)             |
 | `pause_mode`                | `auto`         | Pause mode (`tts`, `manual`, `auto`) |
+| `enable_short_sentence`     | `None`         | Handle short sentences               |
 | `announce_chapters`         | `true`         | Speak chapter titles                 |
 | `chapter_pause_after_title` | `2.0`          | Pause after chapter title            |
 | `phonemization_lang`        | `None`         | Override phonemization language      |

{ttsforge-0.1.1 → ttsforge-0.1.2}/README.md

@@ -554,6 +554,7 @@ ttsforge convert book.epub --gpu
 | `pause_paragraph`           | `0.9`          | Paragraph pause (seconds)            |
 | `pause_variance`            | `0.05`         | Pause variance (seconds)             |
 | `pause_mode`                | `auto`         | Pause mode (`tts`, `manual`, `auto`) |
+| `enable_short_sentence`     | `None`         | Handle short sentences               |
 | `announce_chapters`         | `true`         | Speak chapter titles                 |
 | `chapter_pause_after_title` | `2.0`          | Pause after chapter title            |
 | `phonemization_lang`        | `None`         | Override phonemization language      |

{ttsforge-0.1.1 → ttsforge-0.1.2}/docs/cli.rst

@@ -85,6 +85,9 @@ Options
 ``--pause-mode MODE``
    Pause mode: ``tts``, ``manual``, or ``auto``. Default: ``auto``.
 
+``--enable-short-sentence``
+   Enable special handling for short sentences (less than 5 words).
+
 ``--announce-chapters / --no-announce-chapters``
    Read chapter titles aloud before chapter content. Default: enabled.
 

{ttsforge-0.1.1 → ttsforge-0.1.2}/docs/configuration.rst

@@ -452,6 +452,7 @@ Here's an example ``config.json`` with custom settings:
      "pause_paragraph": 0.9,
      "pause_variance": 0.05,
      "pause_mode": "auto",
+     "enable_short_sentence": None,
      "announce_chapters": true,
      "chapter_pause_after_title": 2.0,
       "save_chapters_separately": false,

{ttsforge-0.1.1 → ttsforge-0.1.2}/tests/test_constants.py

@@ -20,7 +20,7 @@ class TestLanguageDescriptions:
 
     def test_all_language_codes_have_descriptions(self):
         """All language codes should have descriptions."""
-        expected_codes = {"a", "b", "e", "f", "h", "i", "j", "p", "z"}
+        expected_codes = {"a", "b", "d", "e", "f", "h", "i", "j", "p", "z"}
         assert set(LANGUAGE_DESCRIPTIONS.keys()) == expected_codes
 
     def test_english_variants(self):
@@ -132,11 +132,6 @@ class TestDefaultVoiceForLang:
                 lang in DEFAULT_VOICE_FOR_LANG
             ), f"Language {lang} needs default voice"
 
-    def test_default_voices_exist_in_voices_list(self):
-        """All default voices should exist in VOICES list."""
-        for lang, voice in DEFAULT_VOICE_FOR_LANG.items():
-            assert voice in VOICES, f"Default voice {voice} for {lang} not in VOICES"
-
     def test_default_voices_match_language(self):
         """Default voices should match their language."""
         for lang, voice in DEFAULT_VOICE_FOR_LANG.items():

{ttsforge-0.1.1 → ttsforge-0.1.2}/ttsforge/__init__.py

@@ -18,7 +18,7 @@ from pykokoro.tokenizer import (
     Tokenizer,
 )
 from pykokoro.constants import SUPPORTED_LANGUAGES
-
+from pykokoro.onnx_backend import VOICE_NAMES_BY_VARIANT
 from .constants import (
     DEFAULT_CONFIG,
     LANGUAGE_DESCRIPTIONS,
@@ -27,23 +27,7 @@ from .constants import (
 )
 
 # Import from pykokoro
-try:
-    from pykokoro.constants import SAMPLE_RATE
-    from pykokoro.onnx_backend import LANG_CODE_TO_ONNX
-except ImportError:
-    # Fallback values if pykokoro not installed
-    SAMPLE_RATE = 24000
-    LANG_CODE_TO_ONNX = {
-        "a": "en-us",
-        "b": "en-gb",
-        "e": "es",
-        "f": "fr-fr",
-        "h": "hi",
-        "i": "it",
-        "j": "ja",
-        "p": "pt",
-        "z": "zh",
-    }
+from pykokoro.constants import SAMPLE_RATE
 
 from .conversion import (
     Chapter,
@@ -73,6 +57,7 @@ __all__ = [
     "LANGUAGE_DESCRIPTIONS",
     "SUPPORTED_OUTPUT_FORMATS",
     "VOICES",
+    "VOICE_NAMES_BY_VARIANT",
     # Conversion
     "Chapter",
     "ConversionOptions",

{ttsforge-0.1.1 → ttsforge-0.1.2}/ttsforge/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.1.1'
-__version_tuple__ = version_tuple = (0, 1, 1)
+__version__ = version = '0.1.2'
+__version_tuple__ = version_tuple = (0, 1, 2)
 
-__commit_id__ = commit_id = 'g08367e850'
+__commit_id__ = commit_id = 'gb31ed0898'

{ttsforge-0.1.1 → ttsforge-0.1.2}/ttsforge/cli/commands_conversion.py

@@ -17,6 +17,7 @@ from typing import Literal, TypedDict, cast
 
 import click
 import numpy as np
+from pykokoro.onnx_backend import DEFAULT_MODEL_QUALITY, ModelQuality
 from rich.panel import Panel
 from rich.progress import (
     BarColumn,
@@ -37,7 +38,6 @@ from ..constants import (
     LANGUAGE_DESCRIPTIONS,
     SUPPORTED_OUTPUT_FORMATS,
     VOICE_PREFIX_TO_LANG,
-    VOICES,
 )
 from ..conversion import (
     Chapter,
@@ -54,6 +54,7 @@ from ..utils import (
     load_config,
     resolve_conversion_defaults,
 )
+from .commands_utility import _resolve_model_source_and_variant, _resolve_voice_names
 from .helpers import DEFAULT_SAMPLE_TEXT, console, parse_voice_parameter
 
 
@@ -64,6 +65,14 @@ class ContentItem(TypedDict):
     page_number: NotRequired[int]
 
 
+def get_voices() -> list[str]:
+    """Get the list of available voices."""
+    cfg = load_config()
+
+    model_source, model_variant = _resolve_model_source_and_variant(cfg)
+    return _resolve_voice_names(model_source, model_variant)
+
+
 @click.command()
 @click.argument("epub_file", type=click.Path(exists=True, path_type=Path))
 @click.option(
@@ -82,7 +91,7 @@ class ContentItem(TypedDict):
 @click.option(
     "-v",
     "--voice",
-    type=click.Choice(VOICES),
+    type=click.Choice(get_voices()),
     help="Voice to use for TTS.",
 )
 @click.option(
@@ -150,6 +159,12 @@ class ContentItem(TypedDict):
     default=None,
     help="Pause mode: 'tts', 'manual', or 'auto' (default: auto).",
 )
+@click.option(
+    "--enable-short-sentence/--disable-short-sentence",
+    "enable_short_sentence",
+    default=None,
+    help="Enable/disable special handling for short sentences.",
+)
 @click.option(
     "--announce-chapters/--no-announce-chapters",
     "announce_chapters",
@@ -296,6 +311,7 @@ def convert(  # noqa: C901
     pause_paragraph: float | None,
     pause_variance: float | None,
     pause_mode: str | None,
+    enable_short_sentence: bool | None,
     announce_chapters: bool | None,
     chapter_pause: float | None,
     title: str | None,
@@ -325,6 +341,10 @@ def convert(  # noqa: C901
     config = load_config()
     model_path = ctx.obj.get("model_path") if ctx.obj else None
     voices_path = ctx.obj.get("voices_path") if ctx.obj else None
+    model_source, model_variant = _resolve_model_source_and_variant(config)
+    model_quality = cast(
+        ModelQuality, config.get("model_quality", DEFAULT_MODEL_QUALITY)
+    )
 
     # Get format first (needed for output path construction)
     fmt = output_format or config.get("default_format", "m4b")
@@ -467,6 +487,9 @@ def convert(  # noqa: C901
         language=language or "a",
         speed=speed or config.get("default_speed", 1.0),
         use_gpu=use_gpu if use_gpu is not None else config.get("use_gpu", False),
+        model_source=model_source,
+        model_variant=model_variant,
+        model_quality=model_quality,
         num_chapters=len(selected_indices) if selected_indices else len(epub_chapters),
         title=effective_title,
         author=effective_author,
@@ -510,6 +533,9 @@ def convert(  # noqa: C901
         output_format=output_format or config.get("default_format", "m4b"),
         output_dir=output.parent,
         use_gpu=use_gpu if use_gpu is not None else config.get("use_gpu", False),
+        model_quality=model_quality,
+        model_source=model_source,
+        model_variant=model_variant,
         silence_between_chapters=silence or config.get("silence_between_chapters", 2.0),
         lang=lang or config.get("phonemization_lang"),
         use_mixed_language=(
@@ -556,6 +582,11 @@ def convert(  # noqa: C901
         pause_mode=(
             pause_mode if pause_mode is not None else config.get("pause_mode", "auto")
         ),
+        enable_short_sentence=(
+            enable_short_sentence
+            if enable_short_sentence is not None
+            else config.get("enable_short_sentence", None)
+        ),
         announce_chapters=(
             announce_chapters
             if announce_chapters is not None
@@ -947,6 +978,10 @@ def sample(
 
     # Load config for defaults
     user_config = load_config()
+    model_source, model_variant = _resolve_model_source_and_variant(user_config)
+    model_quality = cast(
+        ModelQuality, user_config.get("model_quality", DEFAULT_MODEL_QUALITY)
+    )
     resolved_defaults = resolve_conversion_defaults(
         user_config,
         {
@@ -980,6 +1015,9 @@ def sample(
         use_gpu=resolved_defaults["use_gpu"],
         split_mode=resolved_defaults["split_mode"],
         lang=resolved_defaults["lang"],
+        model_quality=model_quality,
+        model_source=model_source,
+        model_variant=model_variant,
         use_mixed_language=(
             use_mixed_language or user_config.get("use_mixed_language", False)
         ),
@@ -1117,6 +1155,9 @@ def _show_conversion_summary(
     language: str,
     speed: float,
     use_gpu: bool,
+    model_source: str,
+    model_variant: str,
+    model_quality: str | None,
     num_chapters: int,
     title: str,
     author: str,
@@ -1139,6 +1180,9 @@ def _show_conversion_summary(
     table.add_row("Chapters", str(num_chapters))
     table.add_row("Voice", voice)
     table.add_row("Language", LANGUAGE_DESCRIPTIONS.get(language, language))
+    table.add_row("Model Source", model_source)
+    table.add_row("Model Variant", model_variant)
+    table.add_row("Model Quality", str(model_quality))
     if lang:
         table.add_row("Phonemization Lang", f"{lang} (override)")
     if use_mixed_language:
@@ -1167,7 +1211,7 @@ def _show_conversion_summary(
 @click.option(
     "-v",
     "--voice",
-    type=click.Choice(VOICES),
+    type=click.Choice(get_voices()),
     help="TTS voice to use.",
 )
 @click.option(
@@ -1271,6 +1315,11 @@ def _show_conversion_summary(
     default=None,
     help="Trim leading/trailing silence from audio.",
 )
+@click.option(
+    "--enable-short-sentence/--disable-short-sentence",
+    default=None,
+    help="Enable special handling for short sentences.",
+)
 @click.pass_context
 def read(  # noqa: C901
     ctx: click.Context,
@@ -1293,6 +1342,7 @@ def read(  # noqa: C901
     pause_paragraph: float | None,
     pause_variance: float | None,
     pause_mode: str | None,
+    enable_short_sentence: bool | None,
 ) -> None:
     """Read an EPUB or text file aloud with streaming playback.
 
@@ -1340,6 +1390,10 @@ def read(  # noqa: C901
 
     # Load config for defaults
     config = load_config()
+    model_source, model_variant = _resolve_model_source_and_variant(config)
+    model_quality = cast(
+        ModelQuality, config.get("model_quality", DEFAULT_MODEL_QUALITY)
+    )
     resolved_defaults = resolve_conversion_defaults(
         config,
         {
@@ -1389,6 +1443,11 @@ def read(  # noqa: C901
     effective_pause_mode = (
         pause_mode if pause_mode is not None else config.get("pause_mode", "auto")
     )
+    effective_enable_short_sentence = (
+        enable_short_sentence
+        if enable_short_sentence is not None
+        else config.get("enable_short_sentence", None)
+    )
 
     # Get language code for TTS
     espeak_lang = LANG_CODE_TO_ONNX.get(effective_language, "en-us")
@@ -1645,11 +1704,15 @@ def read(  # noqa: C901
             model_path=model_path,
             voices_path=voices_path,
             use_gpu=effective_use_gpu,
+            model_quality=model_quality,
+            model_source=model_source,
+            model_variant=model_variant,
         )
         generation = GenerationConfig(
             speed=effective_speed,
             lang=espeak_lang,
             pause_mode=cast(Literal["tts", "manual", "auto"], effective_pause_mode),
+            enable_short_sentence=effective_enable_short_sentence,
             pause_clause=effective_pause_clause,
             pause_sentence=effective_pause_sentence,
             pause_paragraph=effective_pause_paragraph,
@@ -1658,6 +1721,9 @@ def read(  # noqa: C901
         pipeline_config = PipelineConfig(
             voice=effective_voice,
             generation=generation,
+            model_quality=model_quality,
+            model_source=model_source,
+            model_variant=model_variant,
             model_path=model_path,
             voices_path=voices_path,
         )

{ttsforge-0.1.1 → ttsforge-0.1.2}/ttsforge/cli/commands_phonemes.py

@@ -10,9 +10,10 @@ This module contains commands for working with phonemes and pre-tokenized conten
 import re
 import sys
 from pathlib import Path
-from typing import Any
+from typing import Any, cast
 
 import click
+from pykokoro.onnx_backend import DEFAULT_MODEL_QUALITY, ModelQuality
 from rich.progress import (
     BarColumn,
     Progress,
@@ -37,6 +38,7 @@ from ..utils import (
     format_filename_template,
     load_config,
 )
+from .commands_utility import _resolve_model_source_and_variant
 from .helpers import console, parse_voice_parameter
 
 
@@ -500,6 +502,10 @@ def phonemes_convert(
     config = load_config()
     model_path = ctx.obj.get("model_path") if ctx.obj else None
     voices_path = ctx.obj.get("voices_path") if ctx.obj else None
+    model_source, model_variant = _resolve_model_source_and_variant(config)
+    model_quality = cast(
+        ModelQuality, config.get("model_quality", DEFAULT_MODEL_QUALITY)
+    )
 
     # Get book info and metadata
     book_info = book.get_info()
@@ -599,6 +605,9 @@ def phonemes_convert(
         speed=speed,
         output_format=fmt,
         use_gpu=gpu,
+        model_quality=model_quality,
+        model_source=model_source,
+        model_variant=model_variant,
         silence_between_chapters=silence,
         pause_clause=(
             pause_clause
@@ -834,6 +843,12 @@ def phonemes_preview(
             # Auto-detect if voice is a blend
             parsed_voice, parsed_voice_blend = parse_voice_parameter(voice)
 
+            config = load_config()
+            model_source, model_variant = _resolve_model_source_and_variant(config)
+            model_quality = cast(
+                ModelQuality, config.get("model_quality", DEFAULT_MODEL_QUALITY)
+            )
+
             # Initialize converter
             options = ConversionOptions(
                 phoneme_dictionary_path=str(phoneme_dict) if phoneme_dict else None,
@@ -841,6 +856,9 @@ def phonemes_preview(
                 voice_blend=parsed_voice_blend,
                 language=language,
                 output_format="wav",  # Explicitly set WAV format
+                model_quality=model_quality,
+                model_source=model_source,
+                model_variant=model_variant,
             )
             converter = TTSConverter(options)
 

{ttsforge-0.1.1 → ttsforge-0.1.2}/ttsforge/cli/commands_utility.py

@@ -555,6 +555,14 @@ def _resolve_model_source_and_variant(cfg: dict) -> tuple[ModelSource, ModelVari
     return cast(ModelSource, source), cast(ModelVariant, variant)
 
 
+def _resolve_voice_names(
+    model_source: ModelSource = "huggingface",
+    model_variant: ModelVariant = "v1.0",
+) -> list[str]:
+    """Return the list of voice names for the given model variant."""
+    return VOICE_NAMES_BY_VARIANT.get(model_variant, VOICE_NAMES)
+
+
 def _get_cache_voices_path(
     model_source: ModelSource,
     model_variant: ModelVariant,
@@ -708,7 +716,7 @@ def download(ctx: click.Context, force: bool, quality: str | None) -> None:
 
         # ---- voices
         if model_source == "huggingface":
-            voice_names = VOICE_NAMES_BY_VARIANT.get(model_variant, VOICE_NAMES)
+            voice_names = _resolve_voice_names(model_source, model_variant)
             total_voices = len(voice_names)
             voices_task = progress.add_task(
                 f"Downloading voices (0/{total_voices})...", total=total_voices
@@ -1269,6 +1277,12 @@ def list_names(  # noqa: C901
         )
         console.print("[dim]Type 'q' to quit, 's' to skip, 'r' to replay.[/dim]\n")
 
+        cfg = load_config()
+        model_source, model_variant = _resolve_model_source_and_variant(cfg)
+        model_quality = cast(
+            ModelQuality, cfg.get("model_quality", DEFAULT_MODEL_QUALITY)
+        )
+
         # Initialize converter with phoneme dictionary
         try:
             # Auto-detect if voice is a blend
@@ -1279,6 +1293,9 @@ def list_names(  # noqa: C901
                 voice=parsed_voice or "af_sky",
                 voice_blend=parsed_voice_blend,
                 language=language,
+                model_quality=model_quality,
+                model_source=model_source,
+                model_variant=model_variant,
             )
             converter = TTSConverter(options)
 

{ttsforge-0.1.1 → ttsforge-0.1.2}/ttsforge/cli/helpers.py

@@ -50,6 +50,7 @@ DEFAULT_SAMPLE_TEXT = (
 DEMO_TEXT = {
     "a": "Hello! This audio was generated by {voice}. How do you like it?",
     "b": "Hello! This audio was generated by {voice}. How do you like it?",
+    "d": "Hallo! Dieses Audio wurde von {voice} erzeugt. Wie gefallt es Ihnen?",
     "e": "Hola! Este audio fue generado por {voice}. Que te parece?",
     "f": "Bonjour! Cet audio a ete genere par {voice}. Comment le trouvez-vous?",
     "h": "Namaste! Yah audio {voice} dwara banaya gaya hai. Aapko kaisa laga?",

{ttsforge-0.1.1 → ttsforge-0.1.2}/ttsforge/constants.py

@@ -3,10 +3,10 @@
 # from pykokoro.onnx_backend import VOICE_NAMES_V1_0
 # from pykokoro.onnx_backend import VOICE_NAMES_V1_1_ZH, VOICE_NAMES_V1_1_DE
 
-from pykokoro.onnx_backend import VOICE_NAMES_V1_0 as VOICE_NAMES
+from pykokoro.onnx_backend import DEFAULT_MODEL_SOURCE, VOICE_NAMES_V1_0
 
 # Re-export from pykokoro for convenience
-VOICES = VOICE_NAMES
+VOICES = VOICE_NAMES_V1_0
 
 # Audio constants from pykokoro
 try:
@@ -24,6 +24,7 @@ PROGRAM_DESCRIPTION = "Generate audiobooks from EPUB files using Kokoro ONNX TTS
 LANGUAGE_DESCRIPTIONS = {
     "a": "American English",
     "b": "British English",
+    "d": "German",
     "e": "Spanish",
     "f": "French",
     "h": "Hindi",
@@ -35,6 +36,8 @@ LANGUAGE_DESCRIPTIONS = {
 
 # ISO language code to ttsforge language code mapping
 ISO_TO_LANG_CODE = {
+    "de": "d",
+    "de-de": "d",
     "en": "a",  # Default to American English
     "en-us": "a",
     "en-gb": "b",
@@ -62,6 +65,8 @@ VOICE_PREFIX_TO_LANG = {
     "am": "a",  # American Male
     "bf": "b",  # British Female
     "bm": "b",  # British Male
+    "df": "d",  # German Female
+    "dm": "d",  # German Male
     "ef": "e",  # Spanish Female
     "em": "e",  # Spanish Male
     "ff": "f",  # French Female
@@ -82,6 +87,7 @@ VOICE_PREFIX_TO_LANG = {
 DEFAULT_VOICE_FOR_LANG = {
     "a": "af_heart",
     "b": "bf_emma",
+    "d": "df_eva",
     "e": "ef_dora",
     "f": "ff_siwis",
     "h": "hf_alpha",
@@ -115,6 +121,7 @@ DEFAULT_CONFIG = {
     "use_gpu": False,  # GPU requires onnxruntime-gpu
     # Model quality: fp32, fp16, q8, q8f16, q4, q4f16, uint8, uint8f16
     "model_quality": "fp32",
+    "model_source": DEFAULT_MODEL_SOURCE,
     "model_variant": "v1.0",
     "silence_between_chapters": 2.0,
     "save_chapters_separately": False,
@@ -128,6 +135,7 @@ DEFAULT_CONFIG = {
     "pause_paragraph": 0.9,
     "pause_variance": 0.05,
     "pause_mode": "auto",  # "tts", "manual", or "auto
+    "enable_short_sentence": None,
     # Language override for phonemization (e.g., 'de', 'fr', 'en-us')
     # If None, language is determined from voice prefix
     "phonemization_lang": None,
@@ -154,6 +162,7 @@ AUDIO_CHANNELS = 1
 SAMPLE_TEXTS = {
     "a": "This is a sample of the selected voice.",
     "b": "This is a sample of the selected voice.",
+    "d": "Dies ist ein Beispiel für die ausgewählte Stimme.",
     "e": "Este es una muestra de la voz seleccionada.",
     "f": "Ceci est un exemple de la voix sélectionnée.",
     "h": "यह चयनित आवाज़ का एक नमूना है।",  # noqa: E501