372 lines
16 KiB
Python
Executable File
372 lines
16 KiB
Python
Executable File
import os
|
|
from pathlib import Path
|
|
|
|
|
|
def _csv_env(name: str) -> list[str]:
|
|
value = os.environ.get(name, "")
|
|
if not value:
|
|
return []
|
|
return [item.strip() for item in value.split(",") if item.strip()]
|
|
|
|
|
|
APP_ENV = os.environ.get("APP_ENV", "development").strip().lower()
|
|
ADMIN_TOKEN = os.environ.get("ADMIN_TOKEN", "").strip()
|
|
CORS_ORIGINS = _csv_env("CORS_ORIGINS")
|
|
CORS_ORIGIN_REGEX = os.environ.get("CORS_ORIGIN_REGEX", "").strip()
|
|
|
|
# Base paths — each can be overridden independently via its own env var,
|
|
# or all inherit from MEDIA_DIR when left unset.
|
|
MEDIA_DIR = Path(os.environ.get("MEDIA_DIR", "/app/media"))
|
|
UPLOAD_DIR = Path(os.environ.get("UPLOAD_DIR", str(MEDIA_DIR / "uploads")))
|
|
JOBS_DIR = Path(os.environ.get("JOBS_DIR", str(MEDIA_DIR / "jobs")))
|
|
PROJECTS_DIR = Path(os.environ.get("PROJECTS_DIR", str(MEDIA_DIR / "projects")))
|
|
DB_PATH = Path(os.environ.get("DB_PATH", str(MEDIA_DIR / "db" / "stemengine.db")))
|
|
LRC_SYNC_DIR = Path(os.environ.get("LRC_SYNC_DIR", str(MEDIA_DIR / "lrc_sync")))
|
|
|
|
# BS-RoFormer model directory
|
|
BS_ROFORMER_MODEL_DIR = Path(os.environ.get("BS_ROFORMER_MODEL_DIR", "/opt/bs_roformer_models"))
|
|
|
|
# MelBand RoFormer model directory
|
|
MEL_BAND_MODEL_DIR = Path(os.environ.get("MEL_BAND_MODEL_DIR", "/opt/mel_band_models"))
|
|
|
|
# MDX-Net model directory
|
|
MDXNET_MODEL_DIR = Path(os.environ.get("MDXNET_MODEL_DIR", "/opt/mdxnet_models"))
|
|
|
|
# SCNet XL model directory
|
|
SCNET_MODEL_DIR = Path(os.environ.get("SCNET_MODEL_DIR", "/opt/scnet_models"))
|
|
|
|
# MedleyVox model directory
|
|
MEDLEYVOX_MODEL_DIR = Path(os.environ.get("MEDLEYVOX_MODEL_DIR", "/opt/medleyvox_models"))
|
|
|
|
# Apollo model directory
|
|
APOLLO_MODEL_DIR = Path(os.environ.get("APOLLO_MODEL_DIR", "/opt/apollo_models"))
|
|
|
|
# Bandit model directory
|
|
BANDIT_MODEL_DIR = Path(os.environ.get("BANDIT_MODEL_DIR", "/opt/bandit_models"))
|
|
|
|
# Aspiration model directory
|
|
ASPIRATION_MODEL_DIR = Path(os.environ.get("ASPIRATION_MODEL_DIR", "/opt/aspiration_models"))
|
|
|
|
# SepACap model directory (a cappella multi-singer separation)
|
|
SEPACAP_MODEL_DIR = Path(os.environ.get("SEPACAP_MODEL_DIR", "/opt/sepacap_models"))
|
|
|
|
# SepReformer model directory (2-speaker speech separation)
|
|
SEPREFORMER_MODEL_DIR = Path(os.environ.get("SEPREFORMER_MODEL_DIR", "/opt/sepreformer_models"))
|
|
|
|
# Ensemble mode cache directory
|
|
ENSEMBLE_CACHE_DIR = Path(os.environ.get("ENSEMBLE_CACHE_DIR", "/app/media/ensemble_cache"))
|
|
|
|
# Separation models — Demucs + BS-RoFormer
|
|
DEMUCS_MODELS = [
|
|
{"id": "htdemucs", "label": "Demucs v4", "group": "Demucs"},
|
|
{"id": "htdemucs_ft", "label": "Demucs v4 Fine-tuned", "group": "Demucs"},
|
|
{"id": "mdx_extra_q", "label": "Demucs v3 Extra Quantized","group": "Demucs"},
|
|
{"id": "mdx", "label": "Demucs v3", "group": "Demucs"},
|
|
]
|
|
|
|
BS_ROFORMER_MODELS = [
|
|
{"id": "bs_roformer_4stem", "label": "BS-RoFormer 4-stem", "group": "BS-RoFormer"},
|
|
{"id": "bs_roformer_5stem_guitar", "label": "BS-RoFormer 5-stem (+ Guitar)", "group": "BS-RoFormer"},
|
|
{"id": "bs_roformer_5stem_piano", "label": "BS-RoFormer 5-stem (+ Piano)", "group": "BS-RoFormer"},
|
|
{"id": "bs_roformer_6stem", "label": "BS-RoFormer 6-stem (full)", "group": "BS-RoFormer"},
|
|
]
|
|
|
|
# MelBand RoFormer models
|
|
MEL_BAND_MODELS = [
|
|
{"id": "mel_band_roformer_4stem", "label": "MelBand RoFormer 4-stem", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_vocals", "label": "MelBand Vocals (2-stem)", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_vocals_anvuew", "label": "MelBand Vocals Anvuew", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_dereverb", "label": "MelBand Dereverb", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_dereverb_anvuew", "label": "MelBand Dereverb Anvuew", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_dereverb_room", "label": "MelBand Dereverb Room", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_karaoke", "label": "MelBand Karaoke (Lead/Back)", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_karaoke_gabox", "label": "MelBand Karaoke Gabox", "group": "MelBand RoFormer"},
|
|
{"id": "mel_band_deux_becruily", "label": "MelBand RoFormer Deux Becruily", "group": "MelBand RoFormer"},
|
|
]
|
|
|
|
# BS-RoFormer specialized models
|
|
BS_ROFORMER_SPECIALIZED_MODELS = [
|
|
{"id": "bs_roformer_chorus", "label": "BS-RoFormer Male/Female Chorus", "group": "BS-RoFormer"},
|
|
{"id": "bs_roformer_anvuew", "label": "BS-RoFormer Anvuew (Vocals)", "group": "BS-RoFormer"},
|
|
{"id": "bs_roformer_karaoke_anvuew", "label": "BS-RoFormer Karaoke Anvuew", "group": "BS-RoFormer"},
|
|
]
|
|
|
|
# MDX-Net models (from Ultimate Vocal Remover)
|
|
MDXNET_MODELS = [
|
|
{"id": "mdxnet_voc_ft", "label": "MDX-Net Vocals FT", "group": "MDX-Net"},
|
|
{"id": "mdxnet_inst_hq_1", "label": "MDX-Net Inst HQ 1", "group": "MDX-Net"},
|
|
{"id": "mdxnet_inst_hq_2", "label": "MDX-Net Inst HQ 2", "group": "MDX-Net"},
|
|
{"id": "mdxnet_karaoke", "label": "MDX-Net Karaoke", "group": "MDX-Net"},
|
|
{"id": "mdxnet_dereverb", "label": "MDX-Net Dereverb", "group": "MDX-Net"},
|
|
{"id": "mdxnet_drumsep", "label": "MDX-Net Drum Separation", "group": "MDX-Net"},
|
|
]
|
|
|
|
# SCNet XL models (state-of-the-art 2024-2025)
|
|
SCNET_MODELS = [
|
|
{"id": "scnet_xl", "label": "SCNet XL (2-stem)", "group": "SCNet"},
|
|
{"id": "scnet_xl_ihf", "label": "SCNet XL IHF (2-stem)", "group": "SCNet"},
|
|
{"id": "scnet_large", "label": "SCNet Large (4-stem)", "group": "SCNet"},
|
|
{"id": "scnet_large_starrytong", "label": "SCNet Large Starrytong", "group": "SCNet"},
|
|
]
|
|
|
|
# BS-RoFormer SW (Swin Transformer variant)
|
|
BS_ROFORMER_SW_MODELS = [
|
|
{"id": "bs_roformer_sw_6stem", "label": "BS-RoFormer SW 6-stem", "group": "BS-RoFormer SW"},
|
|
]
|
|
|
|
# MedleyVox models (specialized vocal separation)
|
|
MEDLEYVOX_MODELS = [
|
|
{"id": "medleyvox", "label": "MedleyVox (Vocals)", "group": "MedleyVox"},
|
|
{"id": "medleyvox_55ep", "label": "MedleyVox 55 Epochs", "group": "MedleyVox"},
|
|
{"id": "medleyvox_66ep", "label": "MedleyVox 66 Epochs", "group": "MedleyVox"},
|
|
{"id": "medleyvox_88ep", "label": "MedleyVox 88 Epochs", "group": "MedleyVox"},
|
|
{"id": "medleyvox_100ep", "label": "MedleyVox 100 Epochs", "group": "MedleyVox"},
|
|
]
|
|
|
|
# Apollo models (vocal enhancement/reverb removal)
|
|
APOLLO_MODELS = [
|
|
{"id": "apollo_vocals", "label": "Apollo Vocals (Enhancement)", "group": "Apollo"},
|
|
{"id": "apollo_edm", "label": "Apollo EDM", "group": "Apollo"},
|
|
{"id": "apollo_edm_big", "label": "Apollo EDM Big", "group": "Apollo"},
|
|
]
|
|
|
|
# Aspiration models (breath/aspiration removal)
|
|
ASPIRATION_MODELS = [
|
|
{"id": "aspiration_full", "label": "Aspiration Removal (Full)", "group": "Aspiration"},
|
|
{"id": "aspiration_light", "label": "Aspiration Removal (Light)", "group": "Aspiration"},
|
|
]
|
|
|
|
# Bandit-v2 models (multi-language/multi-stem)
|
|
BANDIT_MODELS = [
|
|
{"id": "bandit_v2_multi", "label": "Bandit v2 Multi-Language", "group": "Bandit"},
|
|
{"id": "bandit_v2_eng", "label": "Bandit v2 English", "group": "Bandit"},
|
|
]
|
|
|
|
# SepACap models (a cappella multi-singer vocal separation -- 7 stems)
|
|
SEPACAP_MODELS = [
|
|
{"id": "sepacap", "label": "SepACap (7-stem A Cappella)", "group": "SepACap"},
|
|
]
|
|
|
|
# SepReformer models (2-speaker speech separation -- NeurIPS 2024)
|
|
SEPREFORMER_MODELS = [
|
|
{"id": "sepreformer_base", "label": "SepReformer Base (Speech, WSJ0)", "group": "SepReformer"},
|
|
{"id": "sepreformer_large_wsj0", "label": "SepReformer Large (Speech, WSJ0)", "group": "SepReformer"},
|
|
{"id": "sepreformer_large_wham", "label": "SepReformer Large (Noisy Speech, WHAM)", "group": "SepReformer"},
|
|
{"id": "sepreformer_large_whamr", "label": "SepReformer Large (Noisy+Reverb, WHAMR)", "group": "SepReformer"},
|
|
]
|
|
|
|
# Ensemble modes (combines multiple models for best quality)
|
|
ENSEMBLE_MODELS = [
|
|
{"id": "ensemble_vocal_best", "label": "Ensemble - Vocal Best Quality", "group": "Ensemble"},
|
|
{"id": "ensemble_4stem", "label": "Ensemble - 4-Stem", "group": "Ensemble"},
|
|
{"id": "ensemble_6stem", "label": "Ensemble - 6-Stem Full", "group": "Ensemble"},
|
|
]
|
|
|
|
ALL_MODELS = (
|
|
DEMUCS_MODELS +
|
|
BS_ROFORMER_MODELS +
|
|
BS_ROFORMER_SW_MODELS +
|
|
BS_ROFORMER_SPECIALIZED_MODELS +
|
|
MEL_BAND_MODELS +
|
|
MDXNET_MODELS +
|
|
SCNET_MODELS +
|
|
MEDLEYVOX_MODELS +
|
|
APOLLO_MODELS +
|
|
ASPIRATION_MODELS +
|
|
BANDIT_MODELS +
|
|
SEPACAP_MODELS +
|
|
SEPREFORMER_MODELS +
|
|
ENSEMBLE_MODELS
|
|
)
|
|
DEFAULT_DEMUCS_MODEL = "htdemucs"
|
|
DEFAULT_MODEL = DEFAULT_DEMUCS_MODEL
|
|
|
|
ALL_MODEL_IDS = {m["id"] for m in ALL_MODELS}
|
|
|
|
# Canonical stem keys for per-instrument model refinement.
|
|
INSTRUMENT_MODEL_KEYS = (
|
|
"vocals",
|
|
"drums",
|
|
"bass",
|
|
"other",
|
|
"guitar",
|
|
"piano",
|
|
)
|
|
|
|
# Output formats
|
|
OUTPUT_FORMATS = [
|
|
{"id": "mp3", "label": "MP3 (256 kbps)"},
|
|
{"id": "mp3_320", "label": "MP3 (320 kbps)"},
|
|
{"id": "flac", "label": "FLAC (lossless)"},
|
|
{"id": "wav", "label": "WAV (lossless)"},
|
|
]
|
|
DEFAULT_OUTPUT_FORMAT = "mp3"
|
|
|
|
# Transcription modes — basic-pitch (in-process) and Omnizart (CLI/sidecar)
|
|
# Each entry carries an "engine" tag so the frontend can group/disable entries
|
|
# and a human-readable "description" surfaced as tooltip text.
|
|
TRANSCRIBE_MODES = [
|
|
{
|
|
"id": "polyphonic",
|
|
"label": "Polyphonic",
|
|
"engine": "basic-pitch",
|
|
"description": "General multi-instrument MIDI — basic-pitch model (fast, in-process)",
|
|
},
|
|
{
|
|
"id": "melody",
|
|
"label": "Melody / Vocal",
|
|
"engine": "basic-pitch",
|
|
"description": "Single-voice / vocal melody extraction — basic-pitch with melodia trick",
|
|
},
|
|
# ── Omnizart modes ──────────────────────────────────────────────────────
|
|
# Require OMNIZART_ENABLED=true; served by the omnizart-sidecar container
|
|
# (or a local Omnizart installation when OMNIZART_SIDECAR_URL is unset).
|
|
{
|
|
"id": "omnizart-music",
|
|
"label": "Omnizart — Multi-instrument",
|
|
"engine": "omnizart",
|
|
"description": "Full-arrangement, instrument-aware MIDI — Omnizart 'music' mode (slow, GPU)",
|
|
},
|
|
{
|
|
"id": "omnizart-vocal",
|
|
"label": "Omnizart — Vocal Melody",
|
|
"engine": "omnizart",
|
|
"description": "Vocal melody extraction — Omnizart 'vocal' mode",
|
|
},
|
|
{
|
|
"id": "omnizart-vocal-contour",
|
|
"label": "Omnizart — Vocal Contour",
|
|
"engine": "omnizart",
|
|
"description": "Vocal pitch contour (F0-based) — Omnizart 'vocal-contour' mode",
|
|
},
|
|
]
|
|
DEFAULT_TRANSCRIBE_MODE = "polyphonic"
|
|
|
|
# Omnizart availability — set OMNIZART_ENABLED=true to unlock the modes above.
|
|
# When False the router returns 503 for any omnizart-* mode request.
|
|
OMNIZART_ENABLED: bool = os.environ.get("OMNIZART_ENABLED", "false").lower() == "true"
|
|
|
|
# When set, omnizart_service will POST audio to this URL and retrieve the MIDI
|
|
# (sidecar HTTP mode). When empty the service falls back to a local subprocess.
|
|
OMNIZART_SIDECAR_URL: str = os.environ.get("OMNIZART_SIDECAR_URL", "").strip()
|
|
|
|
# basic-pitch inference parameter presets per mode
|
|
TRANSCRIBE_PRESETS = {
|
|
"polyphonic": {
|
|
"onset_threshold": 0.5,
|
|
"frame_threshold": 0.3,
|
|
"minimum_note_length": 58.0,
|
|
"melodia_trick": False,
|
|
},
|
|
"melody": {
|
|
"onset_threshold": 0.5,
|
|
"frame_threshold": 0.3,
|
|
"minimum_note_length": 58.0,
|
|
"minimum_frequency": 65.0, # Hz — below typical singing range
|
|
"melodia_trick": True,
|
|
},
|
|
}
|
|
|
|
# Hardware
|
|
CPU_SEPARATION = os.environ.get("CPU_SEPARATION", "true").lower() != "false"
|
|
|
|
# Limits
|
|
MAX_UPLOAD_BYTES = int(os.environ.get("MAX_UPLOAD_MB", "500")) * 1024 * 1024
|
|
POLL_INTERVAL_SECONDS = 5.0
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Presets for stem mixing
|
|
# ---------------------------------------------------------------------------
|
|
|
|
# Available preset types
|
|
PRESET_TYPES = [
|
|
{"id": "stems_only", "label": "Stems Only", "description": "Generate separated stems only"},
|
|
{"id": "static", "label": "Static Mix", "description": "Remove specific elements completely"},
|
|
{"id": "instrumental", "label": "Instrumental", "description": "Remove vocals only"},
|
|
{"id": "dynamic", "label": "Dynamic Mix", "description": "Adjust volume levels of each element"},
|
|
]
|
|
|
|
# Static preset options (what to remove)
|
|
STATIC_OPTIONS = [
|
|
{"id": "no_vocals", "label": "No Vocals", "description": "Remove vocals (keep drums, bass, other)"},
|
|
{"id": "no_drums", "label": "No Drums", "description": "Remove drums (keep vocals, bass, other)"},
|
|
{"id": "no_bass", "label": "No Bass", "description": "Remove bass (keep vocals, drums, other)"},
|
|
{"id": "no_vocals_no_bass", "label": "No Vocals + No Bass", "description": "Remove vocals and bass"},
|
|
{"id": "no_drums_no_bass", "label": "No Drums + No Bass", "description": "Remove drums and bass (vocals + other)"},
|
|
]
|
|
|
|
# Default volume levels for dynamic preset (0.0 to 2.0, where 1.0 = original)
|
|
DEFAULT_DYNAMIC_LEVELS = {
|
|
"vocals": 1.0,
|
|
"drums": 1.0,
|
|
"bass": 1.0,
|
|
"other": 1.0,
|
|
}
|
|
|
|
# Available stems and their labels (varies by model)
|
|
STEM_LABELS = {
|
|
"vocals": "Vocals",
|
|
"drums": "Drums",
|
|
"bass": "Bass",
|
|
"other": "Other",
|
|
"guitar": "Guitar",
|
|
"piano": "Piano",
|
|
"lead": "Lead Vocals",
|
|
"backing": "Backing Vocals",
|
|
"male": "Male Vocals",
|
|
"female": "Female Vocals",
|
|
"instrumental": "Instrumental",
|
|
"dry_vocals": "Dry Vocals",
|
|
"reverb": "Reverb",
|
|
# SepACap a cappella stems
|
|
"alto": "Alto",
|
|
"soprano": "Soprano",
|
|
"tenor": "Tenor",
|
|
"finger_snap": "Finger Snap",
|
|
"vocal_percussion": "Vocal Percussion",
|
|
# SepReformer speech stems
|
|
"speaker_1": "Speaker 1",
|
|
"speaker_2": "Speaker 2",
|
|
}
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# Project folder structure
|
|
# ---------------------------------------------------------------------------
|
|
|
|
PROJECT_SUBFOLDERS = {
|
|
"stems": "Stems", # Always generated - contains separated stems
|
|
"instrumental": "Instrumental", # Contains instrumental mixes (no vocals)
|
|
"static": "Static", # Contains static mixes (specific elements removed)
|
|
"dynamic": "Dynamic", # Contains dynamic mixes (volume adjusted)
|
|
"transcriptions": "Transcriptions", # Contains transcription outputs
|
|
"lrc": "LRC", # Contains synced lyrics files
|
|
}
|
|
|
|
# Subsonic integration
|
|
IMPORTS_DIR = MEDIA_DIR / "imports"
|
|
SUBSONIC_CONFIG_PATH = MEDIA_DIR / "subsonic.json"
|
|
SETTINGS_PATH = MEDIA_DIR / "settings.json"
|
|
|
|
# Lyric transcription (Whisper models)
|
|
WHISPER_MODEL_DIR = Path(os.environ.get("WHISPER_MODEL_DIR", "/opt/whisper_models"))
|
|
DEFAULT_WHISPER_MODEL = os.environ.get("DEFAULT_WHISPER_MODEL", "large-v2")
|
|
|
|
# Lyric transcription (Parakeet self-hosted runtime)
|
|
PARAKEET_ENABLED: bool = os.environ.get("PARAKEET_ENABLED", "false").lower() == "true"
|
|
PARAKEET_SIDECAR_URL: str = os.environ.get("PARAKEET_SIDECAR_URL", "").strip()
|
|
PARAKEET_HEALTH_PATH: str = os.environ.get("PARAKEET_HEALTH_PATH", "/health").strip() or "/health"
|
|
PARAKEET_TRANSCRIBE_PATH: str = os.environ.get("PARAKEET_TRANSCRIBE_PATH", "/transcribe").strip() or "/transcribe"
|
|
PARAKEET_TIMEOUT_SEC: float = float(os.environ.get("PARAKEET_TIMEOUT_SEC", "180"))
|
|
DEFAULT_PARAKEET_MODEL = os.environ.get(
|
|
"DEFAULT_PARAKEET_MODEL",
|
|
"parakeet-1.1b-rnnt-multilingual-asr",
|
|
)
|
|
|
|
# Create directories on import
|
|
UPLOAD_DIR.mkdir(parents=True, exist_ok=True)
|
|
JOBS_DIR.mkdir(parents=True, exist_ok=True)
|
|
PROJECTS_DIR.mkdir(parents=True, exist_ok=True)
|
|
DB_PATH.parent.mkdir(parents=True, exist_ok=True)
|
|
LRC_SYNC_DIR.mkdir(parents=True, exist_ok=True)
|
|
IMPORTS_DIR.mkdir(parents=True, exist_ok=True)
|