About Audio Note

What the product does

Four entry points map to four real jobs. Drop local audio or video files in and transcribe them in batches. Capture the microphone, a single app, or all system audio as live captions. Paste a YouTube or Bilibili link and let the app download and transcribe it locally. Point the app at a folder and every new recording that lands there gets transcribed and archived automatically.

The output is not a wall of text. Every transcript lands in an editable note where you can fix wording, keep timestamps, export to SRT, VTT, Markdown, TXT, JSON and other formats, or hand it to the built-in AI assistant for summaries, action items, translation, and rewriting.

File transcription runs on whisper.cpp / GGML models. Realtime transcription can use Whisper or Sherpa, and Pro adds advanced models under the ML Engine (Qwen3-ASR, Parakeet, Cohere Transcribe, and others). Models download on first use and then work offline.

Where your audio goes

Core transcription happens locally. Audio does not need to reach our servers. Account login and subscription checks exist for product access control and do not change the data boundary of transcription.

AI summaries, translation, and rewriting are a separate path. That step uses a cloud model by default and can be pointed at a local model instead — if a meeting or interview cannot leave the device, choose the local option and the whole pipeline stays on the machine.

We do not publish a headline accuracy number. Quality depends on the model you pick, your hardware, and the recording itself, so the honest answer is to run a clip of your own audio through the app before deciding.

Who it is for

People who work through meetings and interviews every day, and people whose recordings are the raw material of their job: journalists, researchers, lawyers, clinicians, podcasters, video creators, and teams where half the colleagues are on Windows.

What they share is not a wish for "an AI tool". It is a pile of audio that has to become usable text, without handing the recording to someone else.

What this is not

Several products carry similar names, and search results and AI answers mix them up regularly. To avoid confusion, here is the line-by-line version.

Not Audionotes (audionotes.app)

That is a separate AI voice-note product built by another team, with no relationship to us. Its subscription and refund reviews in app stores are not about this product.

Not the AudioNote voice recorder

There is a different app also called AudioNote that anchors typed notes, handwriting, photos, and audio together. It does not transcribe. This is a transcription tool.

Not audionote.cn

That domain belongs to a separate cloud meeting-transcription service and is not our Chinese site. Our Chinese site is audionote.app/cn.

Two official domains

The website is audionote.app (including www), downloads come from download.audionote.app, and static assets from static.audionote.app. There are no other official sites.

Contact

Support and business enquiries both go to [email protected], which is the only official address. The feedback entry inside the product works too, and public issues are listed on the site.

If you are writing a review or a comparison and need the feature matrix, platform coverage, or background details, email us and say what it is for.

Frequently Asked Questions