root
ef9573c294
Per-worker OpenVINO CPU runs the MetricGAN BLSTM mask; SpeechBrain keeps STFT/ISTFT. ENHANCE_INFER=auto with torch fallback. Hop is 0.1 s after isolated last_ms ~6 ms (model+ov+beamform).
162 lines
6.7 KiB
Python
162 lines
6.7 KiB
Python
"""Load configuration from environment / .env file."""
|
|
|
|
import os
|
|
from dataclasses import dataclass, field
|
|
from pathlib import Path
|
|
from typing import Optional
|
|
|
|
|
|
def _load_dotenv(path: Path) -> None:
|
|
if not path.is_file():
|
|
return
|
|
for line in path.read_text().splitlines():
|
|
line = line.strip()
|
|
if not line or line.startswith("#") or "=" not in line:
|
|
continue
|
|
key, _, value = line.partition("=")
|
|
os.environ.setdefault(key.strip(), value.strip())
|
|
|
|
|
|
@dataclass
|
|
class Config:
|
|
# connection
|
|
url: str = ""
|
|
api_key: str = ""
|
|
api_secret: str = ""
|
|
room: str = "cameras"
|
|
participant_prefix: str = "cam"
|
|
# video
|
|
width: int = 640
|
|
height: int = 360
|
|
fps: int = 30
|
|
min_fps: int = 5
|
|
video_bitrate: int = 800_000
|
|
video_codec: str = "h264"
|
|
video_encoder: str = "vaapi" # auto|software|hardware|nvenc|vaapi
|
|
vaapi_device: str = "/dev/dri/renderD129"
|
|
# audio
|
|
audio_rate: int = 16000
|
|
audio_channels: int = 1
|
|
# speechbrain cleanup
|
|
enhance_mode: str = "auto" # auto | force | off
|
|
enhance_model: str = "speechbrain/metricgan-plus-voicebank"
|
|
enhance_chunk_s: float = 1.0
|
|
enhance_hop_s: float = 0.1
|
|
enhance_infer: str = "auto" # auto | openvino | torch
|
|
# SpeechBrain Delay-and-Sum (GCC-PHAT) on C920 stereo before enhance
|
|
beamform_mode: str = "auto" # auto | force | off
|
|
torch_num_threads: int = 1
|
|
beamform_tdoa_every: int = 5
|
|
# misc
|
|
publish_timeout_s: float = 60.0
|
|
log_level: str = "INFO"
|
|
cameras: list[str] = field(default_factory=lambda: ["all"])
|
|
# headless displays (LiveKit subscribe -> GStreamer kmssink)
|
|
display_count: int = 3
|
|
display_identities: list[str] = field(
|
|
default_factory=lambda: ["cam-01", "cam-02", "cam-03"])
|
|
display_connectors: list[str] = field(default_factory=list)
|
|
display_identity: str = "display-wall"
|
|
display_roles: list[str] = field(
|
|
default_factory=lambda: ["grid", "speaker", "screenshare"])
|
|
display_speaker_camera: str = "cam-01" # cam-NN or "active"
|
|
display_speaker_participant: str = "speaker"
|
|
display_grid_cols: int = 5
|
|
display_grid_rows: int = 4
|
|
display_grid_width: int = 1920
|
|
display_grid_height: int = 1080
|
|
display_grid_fps: int = 30
|
|
livekit_public_url: str = ""
|
|
|
|
|
|
def _parse_cameras(spec: str) -> list[str]:
|
|
"""Expand 'all' / '1,3,5' / '0-19' into a list of indices ('' = all)."""
|
|
spec = (spec or "all").strip().lower()
|
|
if spec in ("all", ""):
|
|
return ["all"]
|
|
out = []
|
|
for part in spec.split(","):
|
|
part = part.strip()
|
|
if not part:
|
|
continue
|
|
if "-" in part:
|
|
lo, hi = part.split("-", 1)
|
|
out.extend(str(i) for i in range(int(lo), int(hi) + 1))
|
|
else:
|
|
out.append(part)
|
|
return out
|
|
|
|
|
|
def _parse_list(spec: str) -> list[str]:
|
|
"""Comma-separated names, preserving case (HDMI-A-7, cam-01)."""
|
|
spec = (spec or "").strip()
|
|
if not spec:
|
|
return []
|
|
return [p.strip() for p in spec.split(",") if p.strip()]
|
|
|
|
|
|
def load_config(env_path: Optional[Path] = None) -> Config:
|
|
_load_dotenv(Path(env_path or Path(__file__).parent / ".env"))
|
|
|
|
cfg = Config(
|
|
url=os.environ.get("LIVEKIT_URL", ""),
|
|
api_key=os.environ.get("LIVEKIT_API_KEY", ""),
|
|
api_secret=os.environ.get("LIVEKIT_API_SECRET", ""),
|
|
room=os.environ.get("LIVEKIT_ROOM", "cameras"),
|
|
participant_prefix=os.environ.get("PARTICIPANT_PREFIX", "cam"),
|
|
width=int(os.environ.get("VIDEO_WIDTH", "640")),
|
|
height=int(os.environ.get("VIDEO_HEIGHT", "360")),
|
|
fps=int(os.environ.get("VIDEO_FPS", "30")),
|
|
min_fps=int(os.environ.get("VIDEO_MIN_FPS", "5")),
|
|
video_bitrate=int(os.environ.get("VIDEO_BITRATE", "800000")),
|
|
video_codec=os.environ.get("VIDEO_CODEC", "h264").lower(),
|
|
video_encoder=os.environ.get("VIDEO_ENCODER", "vaapi").lower(),
|
|
vaapi_device=os.environ.get("LIBVA_DRM_DEVICE", "/dev/dri/renderD129"),
|
|
audio_rate=int(os.environ.get("AUDIO_SAMPLE_RATE", "16000")),
|
|
audio_channels=int(os.environ.get("AUDIO_CHANNELS", "1")),
|
|
enhance_mode=os.environ.get("ENHANCE_MODE", "auto").lower(),
|
|
enhance_model=os.environ.get(
|
|
"ENHANCE_MODEL", "speechbrain/metricgan-plus-voicebank"
|
|
),
|
|
enhance_chunk_s=float(os.environ.get("ENHANCE_CHUNK_S", "1.0")),
|
|
enhance_hop_s=float(os.environ.get("ENHANCE_HOP_S", "0.1")),
|
|
enhance_infer=os.environ.get("ENHANCE_INFER", "auto").lower(),
|
|
beamform_mode=os.environ.get("BEAMFORM_MODE", "auto").lower(),
|
|
torch_num_threads=int(os.environ.get("TORCH_NUM_THREADS", "1")),
|
|
beamform_tdoa_every=int(os.environ.get("BEAMFORM_TDOA_EVERY", "5")),
|
|
publish_timeout_s=float(os.environ.get("PUBLISH_TIMEOUT_S", "60")),
|
|
log_level=os.environ.get("LOG_LEVEL", "INFO").upper(),
|
|
cameras=_parse_cameras(os.environ.get("CAMERAS", "all")),
|
|
display_count=int(os.environ.get("DISPLAY_COUNT", "3")),
|
|
display_identities=_parse_list(
|
|
os.environ.get("DISPLAY_IDENTITIES", "cam-01,cam-02,cam-03")),
|
|
display_connectors=_parse_list(os.environ.get("DISPLAY_CONNECTORS", "")),
|
|
display_identity=os.environ.get("DISPLAY_IDENTITY", "display-wall"),
|
|
display_roles=_parse_list(
|
|
os.environ.get("DISPLAY_ROLES", "grid,speaker,screenshare"))
|
|
or ["grid", "speaker", "screenshare"],
|
|
display_speaker_camera=os.environ.get("DISPLAY_SPEAKER_CAMERA", "cam-01"),
|
|
display_speaker_participant=os.environ.get(
|
|
"DISPLAY_SPEAKER_PARTICIPANT", "speaker"),
|
|
display_grid_cols=int(os.environ.get("DISPLAY_GRID_COLS", "5")),
|
|
display_grid_rows=int(os.environ.get("DISPLAY_GRID_ROWS", "4")),
|
|
display_grid_width=int(os.environ.get("DISPLAY_GRID_WIDTH", "1920")),
|
|
display_grid_height=int(os.environ.get("DISPLAY_GRID_HEIGHT", "1080")),
|
|
display_grid_fps=int(os.environ.get("DISPLAY_GRID_FPS", "30")),
|
|
livekit_public_url=os.environ.get("LIVEKIT_PUBLIC_URL", ""),
|
|
)
|
|
|
|
if cfg.enhance_mode not in ("auto", "force", "off"):
|
|
raise ValueError(f"ENHANCE_MODE must be auto|force|off, got {cfg.enhance_mode}")
|
|
if cfg.enhance_infer not in ("auto", "openvino", "torch"):
|
|
raise ValueError(
|
|
f"ENHANCE_INFER must be auto|openvino|torch, got {cfg.enhance_infer}")
|
|
if cfg.beamform_mode not in ("auto", "force", "off"):
|
|
raise ValueError(
|
|
f"BEAMFORM_MODE must be auto|force|off, got {cfg.beamform_mode}")
|
|
if cfg.torch_num_threads < 1:
|
|
raise ValueError("TORCH_NUM_THREADS must be >= 1")
|
|
if cfg.beamform_tdoa_every < 1:
|
|
raise ValueError("BEAMFORM_TDOA_EVERY must be >= 1")
|
|
return cfg
|