SkillAgentSearch skills...

macdub

Real-time offline dubbing for macOS — capture any app, transcribe, translate and speak locally.

Install / Use

claude mcp add lordbasex -- npx -y github:lordbasex/macdub

If the server publishes to npm under a different name, use that package instead — check the repo README.

About this skill
🔌

MCP Server

Model Context Protocol server

Quality Score

81/100

Supported Platforms

Claude Code
Claude Desktop

Tags

Our assessment of macdub

macdub scores 81/100 on our quality scale, 3460th of 4,624 Development & Engineering skills we index.

Its MCP Server is 36 KB long, well organised into 26 sections with 9 code examples: a thorough specification that gives an agent plenty to work with.

It has 10 GitHub stars, so there is little community track record yet; judge it on its content.

Substance
30/30
Structure
20/20
Description
12/15
Adoption
4/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated 41 days ago, so macdub is actively maintained.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 92/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

macdub compared with similar skills

All 4 of these similar skills score higher than macdub; compare them before choosing.

SkillScoreStarsUpdatedFormat
macdub (this skill)by lordbasex811041d agoMCP Server
taste-skillby Leonxlnx100110.1ktodayCLAUDE.md
Agent-Reachby Panniantong100108.9ktodayCLAUDE.md
headroomby headroomlabs-ai10078.9ktodayCLAUDE.md
ai-job-searchby ai-job-search-dev10069.6k1d agoCLAUDE.md

Frequently asked questions

How do I install macdub?
Run claude mcp add lordbasex -- npx -y github:lordbasex/macdub. The install tabs above show the steps for each supported agent.
Which AI agents does macdub work with?
It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
Is macdub safe to use?
It is MIT-licensed and scores 92/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is macdub still maintained?
The repository was last updated 41 days ago, so macdub is actively maintained.

MacDub

<p align="center"><img src="graphics/logo.svg" width="128" height="128" alt="MacDub logo: a speech bubble with a waveform on an indigo squircle"></p>

Real-time, fully offline dubbing for macOS.

MacDub captures the audio of any app (Zoom, Teams, VLC, Safari, Chrome, Music — or the whole system), transcribes it, translates it and reads the translation aloud with a system voice, keeping the original quietly in the background like a documentary. Everything runs on your Mac with Apple's own frameworks. No cloud, no API keys, no virtual audio drivers, and the microphone is never touched.

MacDub pipeline: app or system audio → on-device speech recognition → on-device translation → system voice

Screenshots

A tour of MacDub: Dubbing, Live translation, Subtitles, History with playback, AI & MCP, the menu bar panel, the floating subtitle bar and every Settings tab

The screenshots below were taken while MacDub dubbed Apple's WWDC25 session Bring advanced speech-to-text to your app with SpeechAnalyzer playing in Google Chrome, English → Spanish.

| | | |---|---| | Dubbing — pick the app, the languages and a voice, press Start Dubbing screen | Live transcript — original + translation, the word being spoken highlighted, latency per sentence Live transcript while dubbing | | Floating subtitle bar — always on top of the video you are watching Floating subtitle bar | Menu bar panel — status, last lines, Start/Stop and quick controls Menu bar panel | | Live translation (beta) — call app, microphone, both languages, a voice for each side, audio and/or chat Live translation setup | Live conversation — you and them side by side, a "…" bubble while someone speaks, transcription · audio · total per sentence Live translation conversation | | History › Dubbing — recorded audio with a live spectrum, the transcript following karaoke-style, playback modes, transcription · audio · total times per sentence, right-click to copy a line History with player | History › Live translation — each conversation as a chat, both sides' audio in one take, times per sentence, right-click to copy a line History of a live translation conversation | | Subtitles — transcript options, floating bar and exports Subtitles screen | AI & MCP — local summaries and one-click MCP registration AI & MCP screen | | Settings — General, Capture, Speech, Voice, Subtitles, AI & MCP, Extensions, Permissions Settings window | Settings › Voice — quality badges and your own Personal Voice Voice settings with Personal Voice | | Settings › Extensions — the Google Meet extension for live translation, one click to install Extensions settings | |

  • macOS 15 Sequoia or later (developed on 15.7.3, also built and run on macOS 26.7 / Apple M1).
  • Universal binary — Intel and Apple Silicon.
  • Builds without Xcode: the Command Line Tools are enough.
  • Interface in English, Spanish and Portuguese (Brazil); easy to add more.
  • MIT licensed. Repository: https://github.com/lordbasex/macdub
  • Created by Federico Pereira lord.basex@gmail.com

Features

| Area | Feature | Details | |---|---|---| | Capture | One app or the entire system | Per-app capture follows helper processes (Chrome/Electron helpers, WebKit for Safari); "🔊 Entire system" captures everything except MacDub's own voice. Zoom, Teams, VLC, browsers, Music… The microphone is never captured. | | | Original audio in the background | Core Audio process tap engine: the original is removed from the output and re-played at the level you choose (default 25 %), or lowered only while the translated voice speaks — documentary style. | | | ScreenCaptureKit fallback | Used automatically when the app has no audio process yet (no background mix in that mode). | | | Audio buffer for snippets | The last 60 s (configurable in Settings › AI & MCP, 0 = off) stay in memory so an assistant can request a WAV excerpt through MCP; never written to disk unless asked. | | Recognition | On-device speech recognition | SFSpeechRecognizer with on-device models (macOS 15+). | | | SpeechAnalyzer on macOS 26 | Chosen automatically when the OS offers it; live fallback to SFSpeechRecognizer if it fails; compiled out on older toolchains. MacDub translates and speaks its finalized results — whole, corrected, punctuated sentences (85–92 % of sentences in one piece in the benchmark, against 35 % segmenting its volatile results) — and shows the volatile text live. The engine can be picked in Settings › Capture › Engine, which also says which one was detected and which one is running. | | | Smart sentence segmentation | Cuts on punctuation, clause marks, length, pending time and silence; re-anchors when the recognizer rewrites earlier words. Unit-tested. | | Translation | On-device translation | Apple's Translation framework, 21 languages, models downloaded once. | | Live translation (beta, macOS 26+) | Talk through Meet, Zoom or any call app | What you say reaches the call translated, in the voice you choose (your Personal Voice included), through a virtual microphone (BlackHole, installed from MacDub with one click); what they say reaches your headphones in your language. Two SpeechAnalyzers at once with fast results: about 1.8 s from the end of a sentence to its translated voice. | | | Google Meet extension | Switches Meet's microphone to the translated voice by itself (and back when you stop), posts your translation in the chat and translates the chat messages — Settings › Extensions. | | | Conversation view | Chat-style, with a "…" bubble while someone speaks and, for each sentence, transcription · audio · total times; choose the microphone, the languages (those on your Mac first) and a voice for each side. | | Voice | System voices with quality badges | 🟢 Premium · 🟡 Enhanced · ⚪ Compact · ⚫ novelty; per-voice rate and volume; one-click reload after downloading voices. (Siri voices are not available to third-party apps.) | | | Personal Voice | Dub with your own voice: the Personal Voice recorded in System Settings › Accessibility › Personal Voice appears first in the list (👤) once MacDub is allowed to use it (Settings › Voice › Personal Voice › Allow…). Apple supports it from macOS 14 Sonoma, so every macOS MacDub runs on (15+); creating the voice needs a Mac with Apple silicon, and apps cannot create one — Create in System Settings… opens the pane. It speaks the language it was recorded in. | | | Latency management | Gentle speed-up when behind (configurable or off), stale sentences skipped past Max delay, backlog dropped with one click. Measured: translation ≈ 0.4 s, sentence end → voice ≈ 0.8 s with an empty queue, ≈ 2 s mean. | | | Karaoke highlighting | The word being spoken is highlighted in the panel, the floating bar and the menu bar panel. | | Subtitles | Transcript | While dubbing, the Dubbing screen becomes the live transcript: original + translation, per-sentence latency, level meter, skip/clear. The Subtitles screen holds the transcript options and exports; subtitles-only mode = voice off. | | | Floating subtitle bar (⌘B) | Always-on-top pill: 1–6 lines, adjustable width and text size, drag handle, Stop/Start, quick settings, hide. | | Menu bar | Status icon | Green dot while dubbing, amber while starting; click for a panel with the last lines, Start/Stop and quick controls; right-click for a menu. | | | Menu-bar-only mode | Hide the Dock icon; open at login. | | | Automation | Auto-start when the selected app starts playing audio; silence watchdog (notice or automatic stop). | | Shortcuts | Global | ⌃⌥D start/stop · ⌃⌥S subtitle bar (no Accessibility permission needed). | | | In-app | ⌘R start/stop · ⌘B subtitle bar · ⌘K skip backlog · ⌘L clear (starts a new session) · ⌘E export .srt · ⇧⌘E export .md · ⌘Y history · ⌘, settings. | | Transcripts | Export | .srt (translation / original / both), .md with a session header (made for LLMs), .txt. File names carry no spaces (macdub-transcript-<date>.<ext>). | | | Session history (⌘Y) | Every session archived when it stops, with the original audio recorded as AAC (~/.macdub/audio). Play it back with a live spectrum while the transcript follows karaoke-style — original audio only, documentary style (original lowered under the translated voice) or translated voice only, with original / translated / both texts, remembered between sessions; export audio + .srt under one name (macdub-audio-<date>.m4a/.srt) so VLC picks the subtitles up; delete a session and its audio goes with it. Settings › General shows the size of ~/.macdub and offers Delete all sessions and Factory reset. | | AI | Summaries | Claude Code and Codex CLIs, Ollama and LM Studio (choose the model), Apple Intelligence on macOS 26, or paste into Claude Desktop / ChatGPT — in the dubbing target language. A live seconds counter runs while it works; the summary shows how long it took and the tokens used (input/output; exact for Apple Intelligence on 26.4+, Claude Code, Ollama and LM Studio, ≈ estimated for Codex) plus the cost for Claude Code. | | | MCP server | 18 tools (including History: list/delete sessions, export audio + .srt, storage usage), live resources with push notifications, saved sessions as resources, 3 prompts; stdio and Streamable HTTP transports; server-side summaries with local models; one-click registration for Claude Code, Claude Desktop and Codex that survives moving the app. | | Platform | Requirements | macOS 15 Sequoia or later (tested on 15.7.3 and 26.7), Intel and Apple Silicon (universal binary). | | | System status | Settings › General › Status: chip, cores (performance + efficiency), memory, macOS version, the speech engine in use and Apple Intelligence availability (with the reason when it is off). | | | Build | No Xcode needed — Command Line Tools with Swift 6; make setup installs them if missing. No third-party dependencies. | | | Languages | Interface in English, Spanish and Portuguese (Brazil); adding one is copying a folder. | | | Quality | Unit tests (Swift Testing), CI on GitHub Actions, release script with notarization and Homebrew cask. | | | License | MIT. |


Install

Homebrew

brew tap lordbasex/macdub https://github.com/lordbasex/macdub
brew install --cask macdub

Download

Grab MacDub-<version>.dmg from the Releases page, open it and drag MacDub to the Applications folder shown next to it (a .zip with the bare app is there too). Until releases are notarized, right-click › Open the first time.

Build from source (no Xcode needed)

git clone https://github.com/lordbasex/macdub.git
cd macdub
make setup      # checks macOS 15+, Command Line Tools with Swift 6, SDK frameworks; creates the dev signing cert
make check      # quick type-check (debug build, no bundle)
make run        # native-arch build → build/MacDub.app, then launches it
make            # universal (x86_64 + arm64) build
make zip        # universal zip to try on another Mac
make dmg        # universal build packed as a drag-to-Applications disk image (build/MacDub-<version>.dmg)
make test       # unit tests

make dmg needs only macOS tools (hdiutil, SetFile, osascript); the mounted image shows the MacDub icon, a background with an arrow, the app and an Applications shortcut. The Finder layout ste

Truncated for display — read the full file on GitHub.

Related Skills

View on GitHub
GitHub Stars10
CategoryDevelopment
Updated1mo ago
Forks0

Languages

Swift

Trust signals

92/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

1 low1 info