Kuulo vs Whisper Notes (and Aiko): great transcription — now what?

Whisper Notes costs $6.99. One time. It runs OpenAI's Whisper model directly on your iPhone, producing transcripts in over 100 languages with accuracy that rivals cloud services, entirely offline. No subscription, no data uploaded, no account required. For $6.99, it is one of the best value applications in the App Store.

Aiko is free. It does the same thing — on-device Whisper transcription, offline, excellent accuracy, file import for existing recordings. Sindre Sorhus built it as an open-source contribution to the on-device AI space.

Both tools do exactly what they say they do: transcribe audio to text, on your device, without internet. If that's the complete job, both are excellent.

The problem is that transcription is rarely the complete job.

What Whisper Notes and Aiko do exceptionally well

The core capability is genuinely impressive and worth acknowledging fully before discussing what's missing.

Offline transcription at cloud-quality accuracy. Whisper Large V3 — the model that powers both Whisper Notes and Aiko — is one of the highest-accuracy speech recognition models available. Running it on-device via Apple's Core ML framework, on the Neural Engine of an A-series or M-series chip, produces transcripts that match or exceed the accuracy of major cloud services for clear speech. The offline quality bar has converged with cloud quality on current iPhone hardware.

100+ language support. Whisper's multilingual training covers over 100 languages. Both apps inherit this. For users who need transcription in languages that cloud services don't support well, on-device Whisper is often the better choice on accuracy alone.

One-time purchase. No subscription, no monthly charge, no credits to manage. $6.99 for Whisper Notes, £0 for Aiko. The economics of on-device AI — no server costs — enable a pricing model that cloud services structurally cannot match.

File import. Both apps accept audio file imports — M4A, MP3, WAV, and other formats. An existing recording from another app or a dedicated voice recorder can be transcribed through either tool. Aiko specifically is excellent for this: drag the file in, wait, get a transcript.

True privacy. Neither app collects data, requires an account, or sends anything anywhere. Aiko's privacy label is clean to the point of being minimal. Whisper Notes is the same.

For a user who needs a transcript of a voice recording, quickly, offline, with no account and no cloud exposure: Whisper Notes or Aiko is the right tool. The price is right. The quality is right. The privacy is right.

What happens after the transcript

The transcript is the beginning of the workflow, not the end. Here is what a Whisper Notes or Aiko transcript gives you:

A wall of text. Accurate text. But undifferentiated, unstructured, unattributed text.

If your recording was a 90-minute lecture, you now have a 15,000-word document that requires reading in full to extract the 800 words of actually important content. The effort saved on transcription has been partially shifted to summary extraction.

If your recording had two speakers — an interviewer and a respondent, a consultant and a patient, a professor and a student — neither app tells you which words belong to which voice. You have the conversation; you don't have the conversation broken down by participant.

If your goal was a SOAP note, a meeting minutes document, or a structured lecture summary, neither app produces it. You have raw material; the structured output is still your job.

This is not a criticism. Whisper Notes and Aiko do what they say they do. The limitation is what they don't do — and for users who need those capabilities, the gap is the entire difference.

The missing layer

The capabilities that sit on top of accurate transcription:

AI summarization. Taking 90 minutes of accurate transcript and producing a 10-bullet summary of key points, decisions, or concepts. This requires a language model running inference on the transcript — a different AI capability from speech recognition, and one that is computationally more demanding.

Speaker diarization. Identifying which voice said which words and labelling the transcript accordingly. "Speaker 1: What brings you in today? / Speaker 2: I've been having this chest pain for three weeks..." — the format that makes a clinical or research transcript analytically useful.

Template-structured output. Generating a SOAP note, meeting minutes, a lecture summary in a specific academic format, or a structured handover note from a transcript requires understanding the template structure and extracting the relevant information from the recording into the right sections.

Live translation. Translating spoken audio from one language to another in real time — a capability distinct from after-the-fact transcript translation.

None of these come with Whisper Notes or Aiko. All of them come with Kuulo — running on-device, offline, on the same Neural Engine that handles the transcription.

The comparison

The natural upgrade path

Whisper Notes and Aiko users are the most natural upgrade candidates for Kuulo. They have already made the key decision: on-device, offline, private transcription is the right architecture for their use case. They understand why audio should stay on the device. They value the one-time pricing model over monthly subscriptions.

What they've found is that a wall of accurate transcript text requires a lot of downstream work — reading, restructuring, extracting, summarizing — that they're doing manually every time. Kuulo does that work on-device.

The specific upgrade trigger: the first time a Whisper Notes user finishes transcribing a 90-minute recording and then spends 45 minutes reading through it trying to extract the key points, they are doing work that AI can do in 90 seconds. The wall of text is the symptom; the missing intelligence layer is the problem.

Honest advice

If you genuinely only need transcription — you process the output manually, you don't need summaries or diarization, and $6.99 or £0 is the right price — Whisper Notes and Aiko are excellent tools. Use them.

If you need the full workflow — transcription that then produces a structured, attributed, summarised output ready to use — Kuulo is the tool for the job. The transcription quality is comparable. The intelligence layer on top is what makes the difference.

Both approaches use the same core speech recognition model. The difference is what happens to the transcript after it's generated — and whether that work happens automatically or falls back to you.