16 projects
best-engine-ai-helper
Pick and pull the best local LLM/VLM for the current hardware.
vocal-helper
Vocal Helper — async producer/consumer pipeline turning a live PCM stream into diarized, transcribed utterances and (optionally) a rolling LLM summary. Stages: Voice Activity Detection → online speaker diarization (Hungarian-style cosine clustering on backend embeddings) → STT (pywhispercpp turbo) → optional analyst (a local model served via Ollama or vLLM, resolved by best-engine-ai-helper from a committed brief).
podcast-helper
Podcast Helper — universal audio stream consumer exposed as library, argparse CLI, click CLI, and FastAPI HTTP surface. URL-in → PCM-out for local files, direct audio URLs, RSS feed enclosures, and yt-dlp-supported sources (YouTube, Vimeo, Twitch, SoundCloud, …). Built on youtube-helper + ffmpeg + feedparser + podcastparser; Shannon-correct resampling, mono downmix or native channels preserved.
standpoint
Know where each option stands: a labelled 2D positioning map from any comparison table.
md2star
Markdown → DOCX/PPTX/PDF bridge built on Pandoc, with curated styling, Mermaid rendering, and bibliography support.
youtube-helper
YouTube Helper — yt-dlp wrapper for downloading videos / audio / thumbnails, resolving direct media URLs, browsing video stream catalogs, and pulling no-API engagement metadata (channel / video / comments / subtitles) on YouTube, Vimeo, DailyMotion, Twitch VOD, SoundCloud, and any yt-dlp-supported source.
elbow-helper
Noise-robust knee/elbow detection: a conservative wrapper around a from-scratch locator that either returns a knee with uncertainty or explicitly abstains.
ann-router
Route to the right approximate-nearest-neighbour backend from measured problem criteria (exact/turbovec/HNSW/FAISS/Annoy/Qdrant/pgvector), with a justified, discussable rationale.
capture-helper
Capture Helper — local-first, library-shaped camera / microphone capture layer for the AI Helpers stack, with a live multi-source scene configurator GUI. Multi-surface: library + argparse CLI + click CLI + FastAPI HTTP surface (+ browser GUI at /gui).
video-helper
Video Helper — utility functions for video processing exposed as library, argparse CLI, click CLI, FastAPI HTTP surface, and MCP tools. Validation, format conversion, frame extraction (VidGear / PyAV / ffmpeg-pipe backends), dense optical flow (DIS/Farneback/RAFT), HTTP-headers passthrough, exact-size padded output, target-file-size two-pass compression, subtitle conversion (SRT/VTT/CSS), and pipeline primitives (black video, image loop, concat, overlay, audio mux, subtitle burn).
os-helper
Cross-platform Python utility functions for file handling, system operations, string manipulation, folder management, hashing, configuration loading, and timing/profiling primitives — exposed as library, argparse CLI and click CLI.
wallet-helper
Never run the same heavy call twice. Persistent, content-addressed memoization for expensive calls (a paid API request, a slow model, any heavy function), with single-flight so concurrent identical calls collapse into one, across process restarts and, optionally, across processes.
sftp-helper
SFTP Helper — utility functions for working with SFTP servers via the system OpenSSH client, exposed as library, argparse CLI, click CLI, and FastAPI HTTP surface. Strict host-key verification, credentials loader, upload / download (archive-accelerated for directories) / list / stat / remove / exists / remote_tempfile context manager.
audio-helper
Audio Helper — utility functions for processing audio files exposed as library, argparse CLI, click CLI and FastAPI HTTP surface. Load audio, convert formats, separate audio sources (Demucs), split and concatenate audio. Shannon-correct resampling via ffmpeg/libswresample.
bucket-helper
Bucket Helper — utility functions for AWS S3 and any S3-compatible object storage (MinIO, Backblaze B2, DigitalOcean Spaces, Cloudflare R2, Wasabi, etc.) via boto3, exposed as library, argparse CLI, click CLI, and FastAPI HTTP surface. credentials() loader, upload / download / list / delete / exists / make_bucket / remote_tempfile.
speaker-helper
Speaker Helper — professional Speech Synthesis (offline + streaming) over a local engine (Voicebox by default). The inverse of vocal-helper: text in, speech out. Library + CLI + REST API, with voice cloning and a built-in AI evaluation gate; runnable locally (conda + pip) or as a Docker server.
