AI Helpers
June 2026
Since April 2024
Contents
- Introduction
- 🔒 The Promise
- Composed example
- 🧱 Core
- 🔊 Audio & voice
- 🎬 Video & capture
- 🌐 Media acquisition
- 🗄️ Storage & transfer
- 🧩 Misc
Introduction
AI Helpers is a suite of 13 Python libraries, licensed under BSD-3-Clause (same as scikit-learn / numpy / scipy), designed to simplify AI development across audio/video processing, live speech transcription, media acquisition, capture, secure storage, caching, and hardware-aware local-model selection. The suite is organized into groups, so you install only the corner you need: Core (OS Helper), Audio & voice (Audio, Vocal), Video & capture (Video, Capture), Media acquisition (YouTube, Podcast), Storage & transfer (Bucket, SFTP), and Misc (md2star, Wallet, Standpoint, Best Engine AI). Whether you're handling media files, transcribing and diarizing conversations, scraping creator engagement metadata, streaming podcasts to PCM for ASR, capturing camera/microphone/screen inputs, or managing secure transfers and object storage, AI Helpers streamlines repetitive tasks so you can focus on building.
Every bundled helper exposes the same surfaces: a Python API, two CLIs (argparse + click), a FastAPI HTTP server, and an MCP tool set. Where it earns its place, it also adds a minimal browser GUI at /gui. Each also ships as an installable Claude Code / OpenCode skill. The suite is local-first by design: your files, audio, camera/mic and documents are processed on your machine with open-source tooling, with no SaaS, no telemetry, no account. The two remote-transfer helpers, Bucket and SFTP, are the honest exception; their job is to move data to storage / servers you choose.
Install just the group you need — every helper is its own package and
all of them are on PyPI, so you install by name (each pulls in
os-helper automatically):
# 🔊 Audio & voice
pip install audio-helper vocal-helper
# 🎬 Video & capture
pip install video-helper capture-helper
# 🌐 Media acquisition
pip install youtube-helper podcast-helper
# 🗄️ Storage & transfer
pip install bucket-helper sftp-helper
# 🧩 Misc (docs, cache, positioning maps, local-model selection)
pip install md2star wallet-helper standpoint best-engine-ai-helper
# 🧱 Core only (foundation utilities)
pip install os-helper
You still need ffmpeg on PATH for the media helpers (audio / video / youtube / podcast / capture / vocal); brew install ffmpeg on macOS 🍎.
🔒 The Promise
Sovereignty by design, through local-first Open Source. The philosophy is simple: the tools you build AI with should run on your machine, keep your data under your control, and never make you rent access to your own workflow. No SaaS, no telemetry, no account, no cloud lock-in, and every line is BSD-3-Clause, so you can read it, fork it, and trust it.
We hold ourselves to that honestly, and we are explicit about exactly where it does (and does not) fully apply:
- Guaranteed local: os-helper, audio-helper, video-helper, capture-helper, vocal-helper, md2star and wallet-helper run entirely on your machine. Your files, audio, camera/microphone streams, documents and cached results are processed locally (ffmpeg, whisper.cpp, pyannote / NeMo, Pandoc, a local Ollama model) and never leave it.
- Fetches only what you ask for: youtube-helper and podcast-helper must contact the sites and feeds you point them at (there is no local-first way to download a remote video or episode), but they upload nothing about you and keep everything local. A few helpers download a model or template once on first run, then work offline.
- Deliberately not local-first: bucket-helper and sftp-helper exist to move your data to remote storage and servers. Sovereignty here means you choose the endpoint: your own MinIO / SFTP box is sovereign; a third-party cloud is your call, never a hidden default.
How we applied it, concretely. Every bundled helper ships the same surfaces so you own the whole stack: a Python API, two CLIs (argparse + click), a FastAPI HTTP server, an MCP tool set for agents, and (where auditing the output in a browser genuinely helps) a minimal GUI at /gui, served by the same local FastAPI app with no build step, no framework, and no external calls. Each helper is also an installable Claude Code / OpenCode skill and documents its exact trigger conditions in a TRIGGERS.md. The two remote-transfer helpers wear the exception openly instead of pretending otherwise. That is the whole promise: powerful local building blocks, honestly labelled.
Composed example
The helpers are designed to chain. Here the suite turns a YouTube talk into a shareable Word document and PDF: acquisition, speech-to-text, and typesetting, each stage a different helper:
🌐 youtube-helper → 🗣️ vocal-helper → 📄 md2star (md2docx / md2pdf)
import youtube_helper as yth # 🌐 acquisition
import audio_helper as ah # 🔊 decode to PCM
import vocal_helper as voh # 🗣️ speech-to-text (Whisper)
URL = "https://www.youtube.com/watch?v=YE7VzlLtp-4"
# 1) Acquire: pull the talk's audio (16 kHz mono is ideal for ASR).
yth.download_audio(URL, "talk.mp3", target_sample_rate=16000)
# 2) Transcribe: Whisper on the decoded PCM (to_numpy=True → float32 array).
pcm, sr = ah.load_audio("talk.mp3", target_sample_rate=16000, to_mono=True, to_numpy=True)
transcript = voh.transcribe_pcm(pcm, sr, language="en")
# 3) Hand off to md2star: write a titled Markdown file for typesetting.
with open("talk.md", "w", encoding="utf-8") as fh:
fh.write(f"# Talk transcript\n\n_Source: {URL}_\n\n{transcript}\n")# 4) Typeset the transcript as a Word document and a PDF (md2star CLIs).
md2docx talk.md # → talk.docx
md2pdf talk.md # → talk.pdf
Because md2star's md → docx → pdf render is faithful and reversible
(see the md2star round-trip identity),
the resulting documents read straight back to Markdown without losing the transcript's text.
🧱 Core
OS Helper
OS Helper is a Python library that provides utility functions for working with different operating systems. It offers a set of tools to simplify common system operations, file handling, hashing, config and timing: the shared foundation every other helper builds on.
🍏 OS Helper
·
📃 Documentation
🔊 Audio & voice
Audio Helper
Audio Helper is a Python library that provides utility functions for processing audio files. It includes features like loading audio, converting formats, separating audio sources, and splitting and concatenating audio files.
🔈 Audio Helper
·
📃 Documentation
Vocal Helper
Vocal Helper is a Python library that turns audio into diarized, transcribed utterances, and optionally a rolling LLM summary of the conversation. It runs as an async producer/consumer pipeline in two modes: online streaming (live PCM to live transcript) and offline batch (full audio buffer for highest-quality analysis), chaining voice activity detection, speaker diarization, speech-to-text and optional summarization. Built on Silero VAD, whisper.cpp, pyannote / NeMo diarization backends and Ollama-served language models, it exposes CLI, Python, HTTP and MCP interfaces for live meetings, podcasts and voicemail processing.
🗣️ Vocal Helper
·
📃 Documentation
🎬 Video & capture
Video Helper
Video Helper is a Python library that provides utility functions for processing video files: multi-backend frame extraction (VidGear / PyAV / ffmpeg-pipe), loading, converting, and working with subtitle formats.
📺 Video Helper
·
📃 Documentation
Capture Helper
Capture Helper is the live input layer for the AI Helpers stack: cross-platform camera and microphone enumeration and capture, yielding (H, W, 3) BGR frames and PCM that compose with the same downstream contracts as video-helper and podcast-helper. It ships CLI, FastAPI, MCP, and a live multi-source scene configurator GUI at /gui: pick cameras and mics, preview them live in the browser (MJPEG + mic-level meters), arrange a scene, and save/load it as a reusable JSON config. Camera and microphone data never leave your machine.
🎥 Capture Helper
·
📃 Documentation
🌐 Media acquisition
YouTube Helper
YouTube Helper is a Python library that provides utility functions for downloading videos, audio, and thumbnails from platforms like YouTube, Vimeo, DailyMotion, SoundCloud and Twitch using yt-dlp. It also supports post-processing tasks such as converting or merging media files with ffmpeg, browsing video stream catalogs, and pulling no-API engagement metadata (channel / video / comments / subtitles).
📹 YouTube Helper
·
📃 Documentation
Podcast Helper
Podcast Helper is a Python library for universal audio stream consumption: URL-in to PCM-out. It accepts local files, direct audio URLs, RSS / Atom feed URLs (auto-picks the latest episode), and any yt-dlp-supported source (YouTube, Vimeo, SoundCloud, Twitch). Built on youtube-helper + ffmpeg + feedparser + podcastparser, with Shannon-correct resampling and either source-native channels or canonical mono downmix.
🎙️ Podcast Helper
·
📃 Documentation
🗄️ Storage & transfer
Bucket Helper
Bucket Helper is a Python library for AWS S3 and any S3-compatible object storage (MinIO, Backblaze B2, DigitalOcean Spaces, Cloudflare R2, Wasabi) via boto3. Same shape as SFTP Helper: credentials loader, upload / download / delete / exists / list_prefix, and a remote_tempfile context manager for stage-and-share flows.
🪣 Bucket Helper
·
📃 Documentation
SFTP Helper
SFTP Helper is a Python library that provides utility functions for interacting with SFTP servers via Paramiko, with strict host-key verification and a remote_tempfile helper, once you specified your SSH key credentials.
🛜 SFTP Helper
·
📃 Documentation
📄 Documents
Part of the AI Helpers suite — on PyPI, install by name.
md2star
md2star is a Markdown → DOCX / PPTX / PDF bridge built on Pandoc, with curated branded templates, Mermaid diagram rendering and bibliography support. A focused tool (BSD-3-Clause, on PyPI: pip install md2star).
⭐ md2star
·
📃 Documentation
♻️ Cache
Part of the AI Helpers suite — on PyPI, install by name.
Wallet Helper
Wallet Helper makes sure you never run the same heavy call twice: a paid API request, a slow model, any expensive function. Results are persistent and content-addressed (keyed on a file's content or bytes, not just arguments), so a repeat is served from a local store instead of recomputed, even across process restarts. And when two identical calls start at once, single-flight collapses them into one, in-process, or across processes / hosts via a fenced SQLite lease or an optional HTTP dedup server. It is a small local-first toolbox, close in spirit to os-helper, that it builds on (BSD-3-Clause, on PyPI: pip install wallet-helper).
♻️ Wallet Helper
·
📃 Documentation
Standpoint
Standpoint turns a comparison table (options as rows, criteria as columns, numbers in the cells) into a labelled 2D positioning map, a short written analysis, and a YAML of coordinates, in a single command. The method is ordinary PCA (the perceptual-map staple); what Standpoint adds is the hand-work you would otherwise do yourself: it orients the map around a reference option, names the axes in plain words in your table's own language, colours and labels the points, and writes it all out. Local-first: the figure renders through vl-convert, axis names come from a local Ollama model, and --no-llm gives a fully deterministic run with no model at all (BSD-3-Clause, on PyPI: pip install standpoint).
🗺️ Standpoint
·
📃 Documentation
Best Engine AI Helper
Best Engine AI Helper picks (and pulls) the best local LLM or VLM for the machine it runs on. It detects available memory (Apple Silicon unified memory, NVIDIA VRAM, or system RAM), consults a bundled model catalog, and selects the highest-scoring model that fits within a configurable safety headroom, then pulls it via Ollama and writes an env file that downstream projects source to find the chosen model. Cheap, deterministic text_model() / vision_model() resolvers let the rest of the suite ask "which model do I use?" without re-probing the hardware. Local-first: the models run on your machine through Ollama (BSD-3-Clause, on PyPI: pip install best-engine-ai-helper).
🧠 Best Engine AI Helper
·
📃 Documentation