Earnote: A Locally Running Smart Note-Taking Tool with Whisper Transcription and Task Generation

Title

Earnote: A Locally Running Smart Note-Taking Tool with Whisper Transcription and Task Generation

Meta Description

Earnote is a free, open-source, local AI note-taking tool that uses Whisper to transcribe lectures and meetings on iPhone, iPad, or Mac, automatically generating tasks, flashcards, and PDF study handouts. No account, no ads, no subscription, with a focus on privacy and offline availability.

Article Content

With AI note-taking tools proliferating, privacy and subscription models have become pain points for many users. A recently launched open-source product on Product Hunt called Earnote attempts to address these needs with "fully local operation" and "free and open source."

Core Features: A Complete Workflow from Recording to Structured Notes

Earnote positions itself as "lecture and meeting notes, written on your device." Users simply record a lecture, meeting, or their own explanation, and Earnote invokes the Whisper model for speech-to-text conversion, then further generates structured note content including task lists, flashcards, and a PDF study handout.

The entire workflow covers the complete chain from raw audio to review-ready materials. For students and professionals, this means a single recording can simultaneously produce to-do items and study materials, saving the time of manual organization.

Whisper is an automatic speech recognition (ASR) model open-sourced by OpenAI in 2022. It uses an encoder-decoder Transformer architecture and was trained on 680,000 hours of multilingual audio. It offers five sizes ranging from tiny (about 39 million parameters) to large (about 1.55 billion parameters), supporting transcription and translation in nearly 100 languages. Whisper's notable advantage is its robustness to noisy environments and accents, and its open-source weights allow fully offline deployment. However, larger models offer higher accuracy at greater resource cost, so mobile devices typically need to use smaller or quantized versions, which explains the limited experience on older devices mentioned later.

Local-First: Dual Guarantees of Privacy and Offline Availability

Earnote's most prominent feature is its privacy design. According to the official description, speech recognition runs entirely on the iPhone, iPad, or Mac without uploading audio to the cloud. On iPhone 15 Pro and later models, even note generation is completed on-device.

This means users can still use core features without an internet connection, while avoiding the risk of sensitive meeting content leaking. For users who regularly handle confidential information or value privacy, this feature is quite attractive.

On-device AI operation relies on local inference frameworks provided by the vendor. In the Apple ecosystem, Core ML can convert trained models into formats optimized for Apple Silicon and invoke the Neural Engine for acceleration; the Foundation Models framework introduced with iOS 18 and macOS Sequoia further opens up on-device calls to the system's built-in large models. These technologies enable speech recognition and large model inference without an internet connection, but they also impose clear thresholds on chip computing power and memory—the A17 Pro in the iPhone 15 Pro was the first to be capable of running larger on-device models, which is the technical background for Earnote limiting on-device note generation to that model and later ones.

Business Model: No Account, No Ads, No Subscription

Unlike mainstream AI note-taking apps, Earnote explicitly promises "no account, no ads, no subscription." The product is open-sourced under the MIT license, allowing users to freely view, modify, and even distribute the code. On iPhone and iPad, the developer offers an optional one-time Pro purchase as support for development work.

This model is relatively rare in the current AI tools market. In particular, the absence of an account system means users can start using it without submitting any personal information.

The MIT license is one of the most permissive open-source licenses recognized by the OSI. It only requires retaining the copyright notice and license notice, and allows users to freely use, modify, merge, publish, sublicense, and even commercially sell the software without warranty liability. Unlike the "viral" nature of GPL, MIT does not require derivative works to be open-sourced as well, so Earnote's release under MIT means other developers can integrate its code into closed-source products. This choice lowers the barrier to adoption and facilitates community spread, but it also means the original author cannot use the license to restrict commercialization, and the project's sustainability relies more on voluntary support models such as one-time Pro purchases.

Community Feedback and Market Positioning

On Product Hunt, Earnote received 75 upvotes, ranking 10th for the day, categorized under Productivity, Open Source, and Education. Although there were only 3 comments, the ranking shows it has attracted some attention among open-source productivity tools.

From a competitive landscape perspective, Earnote differentiates itself from cloud-based transcription note tools such as Otter.ai and Notta. The latter two rely on cloud processing and offer more powerful collaboration and synchronization features, but require subscriptions and raise privacy concerns. Earnote targets the niche market of "local processing + open source," suitable for individual users sensitive to data sovereignty.

Limitations Analysis

Local operation also brings some limitations. The transcription accuracy and speed of on-device Whisper models are affected by hardware performance, and older devices may have a poor experience. In addition, note generation is only supported on-device on iPhone 15 Pro and newer models; other devices may need to rely on the cloud or have limited functionality.

As an open-source project, Earnote's long-term maintenance and update frequency also depend on the developer community. Currently, information on the Product Hunt page is limited, and actual user experience and stability await verification by more users.

Conclusion

Earnote represents a product philosophy of returning to local processing: in an era where AI capabilities are readily available, it keeps processing on the user's device and earns trust through open source and free access. For users who value privacy, need offline availability, and do not want to be tied to subscriptions, it offers a noteworthy option.

One-line Summary

Earnote is a fully local, free, open-source AI recording and note-taking tool focused on privacy and offline availability.

Paragraph Summary

Earnote is an open-source AI note-taking product on Product Hunt, positioned as "lecture and meeting notes, written on your device." After users record audio, it invokes the Whisper model to perform speech-to-text on-device and generates structured notes including task lists, flashcards, and PDF study handouts. The product emphasizes local-first: speech recognition runs entirely on iPhone, iPad, or Mac, and iPhone 15 Pro and later models also support on-device note generation, balancing privacy and offline use. Its business model is open source under the MIT license, with no account, no ads, and no subscription, offering only an optional one-time Pro purchase on iPhone/iPad. It received 75 upvotes on Product Hunt and ranked 10th for the day, competing differently from cloud tools such as Otter.ai and Notta. Limitations include restricted transcription accuracy and speed on older devices, a high device threshold for note generation, and long-term maintenance of the open-source project depending on the community.