pyproject.toml

[build-system]
requires = ["setuptools>=77.0.0", "wheel"]
build-backend = "setuptools.build_meta"

[project]
name = "abstractvoice"
dynamic = ["version"]
description = "Remote-compatible and local TTS/STT, streaming voice output, and optional voice cloning for AI applications"
readme = "README.md"
requires-python = ">=3.9"
license = "MIT"
license-files = ["LICENSE", "third_party_licenses/*"]
authors = [
    {name = "Laurent-Philippe Albou", email = "contact@abstractcore.ai"}
]
classifiers = [
    "Development Status :: 3 - Alpha",
    "Intended Audience :: Developers",
    "Intended Audience :: Information Technology",
    "Intended Audience :: Science/Research",
    "Operating System :: OS Independent",
    "Programming Language :: Python :: 3",
    "Programming Language :: Python :: 3.9",
    "Programming Language :: Python :: 3.10",
    "Programming Language :: Python :: 3.11",
    "Programming Language :: Python :: 3.12",
    "Topic :: Multimedia :: Sound/Audio",
    "Topic :: Multimedia :: Sound/Audio :: Sound Synthesis",
    "Topic :: Multimedia :: Sound/Audio :: Speech",
    "Topic :: Scientific/Engineering :: Artificial Intelligence",
    "Topic :: Software Development :: Libraries :: Python Modules",
]
dependencies = [
    "numpy>=1.24.0",
    "requests>=2.31.0",
    "appdirs>=1.4.0",
]

[project.optional-dependencies]
# Remote OpenAI/OpenAI-compatible audio adapters use core `requests`.
# These no-op extras are provided so installers can declare intent explicitly:
#   pip install "abstractvoice[openai]"
openai = []
openai-compatible = []
remote = []

# Native Apple Python install profile. This is intentionally a direct macOS
# install profile, not a Docker profile; it keeps the local audio, STT, TTS,
# and voice-engine stack available without NVIDIA-only dependencies.
apple = [
    "piper-tts>=1.2.0",
    "onnxruntime>=1.19.0",
    "faster-whisper>=0.10.0",
    "sounddevice>=0.4.6",
    "webrtcvad>=2.0.10",
    "soundfile>=0.12.1",
    "librosa>=0.10.0",
    "librosa>=0.11.0; python_version >= '3.10'",
    "huggingface_hub>=0.20.0",
    "f5-tts>=1.1.0; python_version >= '3.10'",
    "torch>=2.0.0",
    "torchaudio>=2.0.0; python_version >= '3.10'",
    "torchvision>=0.15.0; python_version >= '3.10'",
    "transformers>=4.55.4,<5; python_version < '3.10'",
    "transformers>=5.4.0; python_version >= '3.10'",
    "accelerate>=1.0.0; python_version >= '3.10'",
    "av>=14.0.0; python_version >= '3.10'",
    "audioread>=3.0.0; python_version >= '3.10'",
    "pillow>=11.0.0; python_version >= '3.10'",
    "safetensors>=0.4.0",
    "safetensors>=0.5.0; python_version >= '3.10'",
    "einops>=0.8.0",
    "sentencepiece>=0.1.99",
    "omnivoice>=0.1.5; python_version >= '3.10'",
    "aec-audio-processing>=1.0.1; python_version >= '3.11'",
]

gpu = [
    "piper-tts>=1.2.0",
    "onnxruntime>=1.19.0",
    "faster-whisper>=0.10.0",
    "sounddevice>=0.4.6",
    "webrtcvad>=2.0.10",
    "soundfile>=0.12.1",
    "librosa>=0.10.0",
    "librosa>=0.11.0; python_version >= '3.10'",
    "huggingface_hub>=0.20.0",
    "f5-tts>=1.1.0; python_version >= '3.10'",
    "torch>=2.0.0",
    "torchaudio>=2.0.0; python_version >= '3.10'",
    "torchvision>=0.15.0; python_version >= '3.10'",
    "transformers>=4.55.4,<5; python_version < '3.10'",
    "transformers>=5.4.0; python_version >= '3.10'",
    "accelerate>=1.0.0; python_version >= '3.10'",
    "av>=14.0.0; python_version >= '3.10'",
    "audioread>=3.0.0; python_version >= '3.10'",
    "pillow>=11.0.0; python_version >= '3.10'",
    "safetensors>=0.4.0",
    "safetensors>=0.5.0; python_version >= '3.10'",
    "einops>=0.8.0",
    "sentencepiece>=0.1.99",
    "omnivoice>=0.1.5; python_version >= '3.10'",
    "aec-audio-processing>=1.0.1; python_version >= '3.11'",
]

# Granular local engine extras remain available for smaller custom installs.
piper = [
    "piper-tts>=1.2.0",
]

supertonic = [
    "onnxruntime>=1.19.0",
]

# Local audio input/output and VAD.
audio-io = [
    "sounddevice>=0.4.6",
    "webrtcvad>=2.0.10",
    "soundfile>=0.12.1",
]

# Audio effects (optional; only needed for speed/pitch processing)
audio-fx = [
    "librosa>=0.10.0",
]

# Voice cloning (optional; research-grade, heavy deps).
cloning = [
    "huggingface_hub>=0.20.0",
    "f5-tts>=1.1.0; python_version >= '3.10'",
    "soundfile>=0.12.1",
]

# Chroma (optional; GPU-heavy end-to-end speech model).
chroma = [
    "huggingface_hub>=0.20.0; python_version >= '3.10'",
    "torch>=2.0.0; python_version >= '3.10'",
    "torchaudio>=2.0.0; python_version >= '3.10'",
    "torchvision>=0.15.0; python_version >= '3.10'",
    "transformers>=5.0.0rc0; python_version >= '3.10'",
    "accelerate>=1.0.0; python_version >= '3.10'",
    "av>=14.0.0; python_version >= '3.10'",
    "librosa>=0.11.0; python_version >= '3.10'",
    "audioread>=3.0.0; python_version >= '3.10'",
    "pillow>=11.0.0; python_version >= '3.10'",
    "safetensors>=0.5.0; python_version >= '3.10'",
    "soundfile>=0.12.1; python_version >= '3.10'",
]

# LongCat-AudioDiT (optional; torch/transformers-based TTS + prompt-audio cloning)
audiodit = [
    "huggingface_hub>=0.20.0",
    "torch>=2.0.0",
    "transformers>=4.55.4,<5; python_version < '3.10'",
    "transformers>=5.4.0; python_version >= '3.10'",
    "safetensors>=0.4.0",
    "einops>=0.8.0",
    "sentencepiece>=0.1.99",
    "soundfile>=0.12.1",
]
qwen3-tts = [
    "huggingface_hub>=0.20.0",
    "torch>=2.0.0",
    "transformers>=5.6.0,<6; python_version >= '3.10'",
    "safetensors>=0.4.0",
    "soundfile>=0.12.1",
    "numpy>=1.24.0",
]

# OmniVoice (optional; omnilingual TTS + voice cloning / design)
omnivoice = [
    # Torch-backed and heavy; treat this as an opt-in extra to avoid affecting
    # default installs.
    "huggingface_hub>=0.20.0; python_version >= '3.10'",
    "omnivoice>=0.1.5; python_version >= '3.10'",
    "torch>=2.0.0; python_version >= '3.10'",
    "torchaudio>=2.0.0; python_version >= '3.10'",
    "transformers>=5.4.0; python_version >= '3.10'",
    "accelerate>=1.0.0; python_version >= '3.10'",
    "soundfile>=0.12.1; python_version >= '3.10'",
]

# Optional AEC (Acoustic Echo Cancellation) for true barge-in.
# This is intentionally opt-in because platform support/toolchain requirements vary.
aec = [
    "aec-audio-processing>=1.0.1; python_version >= '3.11'",
]

# Local Speech-to-Text functionality.
stt = [
    "faster-whisper>=0.10.0",
    "soundfile>=0.12.1",
]

# Local Speech-to-Text functionality (Transformers-based Hugging Face models).
stt-hf = [
    "soundfile>=0.12.1",
    "torch>=2.0.0",
    "transformers>=4.55.4,<5; python_version < '3.10'",
    "transformers>=5.4.0; python_version >= '3.10'",
    "accelerate>=1.0.0; python_version >= '3.10'",
    "huggingface_hub>=0.20.0",
    "safetensors>=0.4.0",
    "safetensors>=0.5.0; python_version >= '3.10'",
    "sentencepiece>=0.1.99",
]

# Local FastAPI web example. Production HTTP audio endpoints live in AbstractCore Server.
web = [
    "fastapi>=0.100.0",
    "uvicorn>=0.23.0",
    "python-multipart>=0.0.9",
]

# Full native Apple profile: local voice stack plus the lightweight local web
# example, excluding NVIDIA/CUDA-specific dependencies.
all-apple = [
    "piper-tts>=1.2.0",
    "onnxruntime>=1.19.0",
    "faster-whisper>=0.10.0",
    "sounddevice>=0.4.6",
    "webrtcvad>=2.0.10",
    "soundfile>=0.12.1",
    "librosa>=0.10.0",
    "librosa>=0.11.0; python_version >= '3.10'",
    "huggingface_hub>=0.20.0",
    "f5-tts>=1.1.0; python_version >= '3.10'",
    "torch>=2.0.0",
    "torchaudio>=2.0.0; python_version >= '3.10'",
    "torchvision>=0.15.0; python_version >= '3.10'",
    "transformers>=4.55.4,<5; python_version < '3.10'",
    "transformers>=5.4.0; python_version >= '3.10'",
    "accelerate>=1.0.0; python_version >= '3.10'",
    "av>=14.0.0; python_version >= '3.10'",
    "audioread>=3.0.0; python_version >= '3.10'",
    "pillow>=11.0.0; python_version >= '3.10'",
    "safetensors>=0.4.0",
    "safetensors>=0.5.0; python_version >= '3.10'",
    "einops>=0.8.0",
    "sentencepiece>=0.1.99",
    "omnivoice>=0.1.5; python_version >= '3.10'",
    "aec-audio-processing>=1.0.1; python_version >= '3.11'",
    "fastapi>=0.100.0",
    "uvicorn>=0.23.0",
    "python-multipart>=0.0.9",
]

all-gpu = [
    "piper-tts>=1.2.0",
    "onnxruntime>=1.19.0",
    "faster-whisper>=0.10.0",
    "sounddevice>=0.4.6",
    "webrtcvad>=2.0.10",
    "soundfile>=0.12.1",
    "librosa>=0.10.0",
    "librosa>=0.11.0; python_version >= '3.10'",
    "huggingface_hub>=0.20.0",
    "f5-tts>=1.1.0; python_version >= '3.10'",
    "torch>=2.0.0",
    "torchaudio>=2.0.0; python_version >= '3.10'",
    "torchvision>=0.15.0; python_version >= '3.10'",
    "transformers>=4.55.4,<5; python_version < '3.10'",
    "transformers>=5.4.0; python_version >= '3.10'",
    "accelerate>=1.0.0; python_version >= '3.10'",
    "av>=14.0.0; python_version >= '3.10'",
    "audioread>=3.0.0; python_version >= '3.10'",
    "pillow>=11.0.0; python_version >= '3.10'",
    "safetensors>=0.4.0",
    "safetensors>=0.5.0; python_version >= '3.10'",
    "einops>=0.8.0",
    "sentencepiece>=0.1.99",
    "omnivoice>=0.1.5; python_version >= '3.10'",
    "aec-audio-processing>=1.0.1; python_version >= '3.11'",
    "fastapi>=0.100.0",
    "uvicorn>=0.23.0",
    "python-multipart>=0.0.9",
]

# CI/test dependencies.
test = [
    "pytest>=7.0.0",
    "httpx>=0.23.0",
    "soundfile>=0.12.1",
    "webrtcvad>=2.0.10",
    "tomli>=1.1.0; python_version < '3.11'",
]

# Documentation dependencies.
docs = [
    "mkdocs>=1.6.0",
    "mkdocs-material>=9.0.0",
]

# Development dependencies
dev = [
    "pytest>=7.0.0",
    "black>=22.0.0",
    "flake8>=5.0.0",
    "mkdocs>=1.6.0",
    "mkdocs-material>=9.0.0",
]

[project.urls]
Repository = "https://github.com/lpalbou/abstractvoice"
Documentation = "https://www.lpalbou.info/AbstractVoice/"

[project.scripts]
abstractvoice = "abstractvoice.examples.voice_cli:main"
abstractvoice-prefetch = "abstractvoice.prefetch:main"

[project.entry-points."abstractcore.capabilities_plugins"]
abstractvoice = "abstractvoice.integrations.abstractcore_plugin:register"

[tool.setuptools.dynamic]
version = {attr = "abstractvoice._version.__version__"}

[tool.setuptools.packages.find]
# Explicit discovery avoids setuptools accidentally treating `audio_samples/`
# (or any other repo folder) as an installable top-level package.
include = ["abstractvoice*"]
exclude = ["audio_samples*", "tests*", "docs*"]

[tool.setuptools.package-data]
# Ship built-in voice profile preset packs (JSON).
abstractvoice = ["assets/**/*.json"]

[tool.pytest.ini_options]
minversion = "7.0"
testpaths = ["tests"]
python_files = ["test_*.py"]
python_functions = ["test_*"]
python_classes = ["Test*"]
pythonpath = ["."]
addopts = "-m \"not integration and not model_download and not slow\""
markers = [
    "integration: integration tests that may require sibling repos or optional engines",
    "model_download: tests that may download or require cached model artifacts",
    "slow: slow tests",
    "basic: lightweight contract/unit tests",
]