--- title: "Capture Per-App Audio on Mac Without Virtual Drivers" description: "Record and stream audio from any individual Mac app - Chrome, Spotify, Zoom - without installing virtual audio drivers. AudioPiper uses Core Audio Taps built into macOS." date: 2026-03-29 author: "Ben Racicot" tags: ["Audio Capture", "macOS", "Core Audio Taps", "Recording", "Streaming", "Privacy"] type: "article" canonical: "https://modelpiper.com/blog/capture-app-audio-mac-no-drivers/" --- # Capture Per-App Audio on Mac Without Virtual Drivers > Record and stream audio from any individual Mac app - Chrome, Spotify, Zoom - without installing virtual audio drivers. AudioPiper uses Core Audio Taps built into macOS. ## TL;DR AudioPiper captures audio from any individual app on your Mac - mic, system audio, or just one app like Chrome or Spotify - without Loopback, Soundflower, or any virtual audio driver. It uses Core Audio Taps built into macOS 14.2+, mixes sources in real time, streams to ModelPiper's AI pipelines at 48kHz, and records to WAV. Free on the Mac App Store. You want to record what Chrome is playing. Or capture Zoom audio alongside your microphone. Or stream a YouTube tutorial into a transcription tool. On a Mac, this has been unreasonably hard for decades. The traditional answer was virtual audio drivers - Soundflower, BlackHole, Loopback. They intercept audio at the system level by installing kernel extensions or audio plugins. They work, but they're fragile. macOS updates break them regularly. They sometimes require restarts. Soundflower was abandoned years ago. BlackHole requires manual routing. Loopback costs $99 and still installs system-level components that Apple keeps restricting further with each release. **macOS 14.2 quietly solved this with Core Audio Taps - a native API that lets apps capture audio from any process without drivers, plugins, or kernel extensions.** ## What are Core Audio Taps? Core Audio Taps are an Apple API introduced in macOS 14.2 (Sonoma). They let an app request the audio output of any other running process directly from the system's audio server. No virtual devices. No routing tricks. No kernel extensions. The key difference: it's scoped. You ask for a specific app's audio and you get exactly that process's output. You can capture Chrome without hearing Spotify. You can capture Zoom without hearing your notification sounds. Or you can request system-wide audio to capture everything. This is fundamentally different from virtual drivers, which create a fake audio device that sits between apps and your speakers and intercepts everything. Core Audio Taps read the audio stream directly from the process you specify. It's cleaner, more reliable, and nothing about your system changes - no new devices appear in Audio MIDI Setup, no routing to reconfigure. ## What is AudioPiper? AudioPiper is a free macOS menu bar app that puts a simple interface on top of Core Audio Taps. Open it, and you see every audio source available: your microphone, system audio, and every app currently producing sound. Toggle on what you want to capture. That's it. AudioPiper mixes your selected sources in real time with per-source volume control. You can record the mix to a WAV file, or stream it live to AI tools for transcription, translation, or analysis. There is nothing to install beyond the app itself. No drivers, no configuration, no restart. ## How do you use AudioPiper? Download AudioPiper from the Mac App Store (free). It lives in your menu bar. **To capture a specific app's audio:** Click the menu bar icon. You'll see a list of every running app that's producing audio. Toggle one on. You're now capturing that app's audio output. **To record multiple sources:** Toggle on your microphone plus one or more app sources. Adjust the per-source volume sliders to balance them - for example, turn up the Zoom audio and turn down your mic if you're mostly interested in what the other person is saying. Hit record to save a WAV file. **To feed audio into AI:** AudioPiper can stream the audio mix to [ToolPiper](https://modelpiper.com) (our local AI engine). From there, you can run it through [voice transcription](/blog/local-voice-transcription-mac), [voice chat](/blog/voice-chat-mac-local-ai), or [live translation](/blog/live-translation-mac-local) - all on your Mac, no audio uploaded anywhere. ## What people actually use this for **Recording a Zoom call with both sides.** You're on a client call and want a record of what was discussed. Toggle on Zoom and your microphone. AudioPiper captures both sides of the conversation in one recording, with separate volume controls so you can balance them. The recording stays on your Mac - no cloud transcription service sees it. **Transcribing a webinar or YouTube video.** A two-hour training session is playing in Chrome. You don't want to take notes manually. Toggle on Chrome in AudioPiper, and the audio streams directly into a local transcription pipeline. You get a full text transcript without the audio ever leaving your machine, and without paying per-minute for a cloud API. **Recording a podcast with reference material.** You're recording yourself talking while a reference clip plays in another app. Toggle on your mic plus the reference app. AudioPiper mixes them. No mixing board, no virtual audio routing - just two toggles. **Getting subtitles for anything.** Any audio playing on your Mac can be captured and fed into real-time transcription. A foreign-language video, an accessibility need, a lecture you want to follow in text - capture the app's audio and transcribe it locally. ## What AudioPiper doesn't do It's worth being clear about the edges. AudioPiper captures what apps send to the audio system. If an app uses DRM-protected audio streams (like some streaming services in certain browsers), the audio may not be available to capture. This is an Apple-level restriction, not an AudioPiper limitation. An app needs to be actively producing audio for it to appear in the source list. You can't pre-select an app before it starts playing - toggle it on once you hear it. AudioPiper records to WAV. If you need MP3, AAC, or other compressed formats, you'll need to convert after recording. WAV is lossless and uncompressed, which is ideal for AI processing and archival, but the files are large (~10 MB per minute of stereo audio). AudioPiper captures audio only. If you need screen recording with audio, see [VisionPiper](/blog/mac-screen-capture-ai-streaming) for the video side. ## A note on permissions Core Audio Taps use the Audio Capture permission - not the Screen Recording permission. There's no purple recording indicator in the menu bar. macOS will prompt you once the first time you use it. This matters because the Screen Recording permission prompt is invasive (it appears for every app that wants to see your screen), and the purple dot is distracting during presentations and screenshares. Audio Capture is a lighter, less disruptive permission. ## Get AudioPiper AudioPiper is free on the [Mac App Store](https://apps.apple.com/app/audiopiper/id6759811522). Requires macOS 14.2 or later and Apple Silicon (M1+). For AI workflows (transcription, translation, voice chat), you'll also want [ToolPiper](https://modelpiper.com) - our local AI engine that runs speech-to-text, language models, and text-to-speech on your Mac. AudioPiper captures the audio; ToolPiper processes it. _This is part of a series on [local-first AI workflows on macOS](/blog/local-first-ai-macos). See also: [Voice Transcription](/blog/local-voice-transcription-mac), [Voice Chat](/blog/voice-chat-mac-local-ai), [Transcribe & Summarize](/blog/transcribe-summarize-mac)._ ## FAQ ### How do I capture audio from a specific app on Mac? Install AudioPiper from the Mac App Store (free). Click the menu bar icon - you'll see every running app that's producing audio listed as a source. Toggle on the one you want. AudioPiper captures that app's output directly using Core Audio Taps, a native macOS API. No virtual audio drivers, no routing configuration. ### Do I need Soundflower, BlackHole, or Loopback? No. Those tools use virtual audio drivers or kernel extensions to intercept audio. AudioPiper uses Core Audio Taps instead - a native macOS API that reads audio directly from any process. There's nothing to install beyond the app itself, and macOS updates won't break your setup. ### Can I capture my microphone and an app's audio at the same time? Yes. Toggle on your microphone and one or more app sources. AudioPiper mixes them with separate volume sliders so you can balance the levels - for example, turn up Zoom audio and turn down your mic if you mainly want to capture the other person. You can record the mix or stream it to AI tools. ### Can I transcribe what's playing in Chrome or Safari? Yes. Toggle on Chrome (or Safari, or any browser) in AudioPiper. The audio streams to [ToolPiper](https://modelpiper.com), where you can run it through [voice transcription](/blog/local-voice-transcription-mac) or [transcribe & summarize](/blog/transcribe-summarize-mac). The audio stays on your Mac - nothing is uploaded to a cloud transcription service. This works for YouTube videos, webinars, online courses, or anything playing in a browser tab. ### Does capturing app audio show the purple recording indicator in the menu bar? No. Core Audio Taps use the Audio Capture permission, which is separate from the Screen Recording permission. There's no purple dot. macOS asks for Audio Capture permission once the first time you use it. ### Can I record for a long time without running out of memory? Yes. AudioPiper uses a streaming writer that keeps memory usage flat regardless of recording length. Whether you record for 5 minutes or 5 hours, memory stays the same. Recordings are saved as WAV files (48kHz, 16-bit, stereo) - roughly 10 MB per minute. ### What macOS version do I need? macOS 14.2 (Sonoma) or later - that's when Apple introduced Core Audio Taps. You also need Apple Silicon (M1 or later). AudioPiper is free on the Mac App Store.