Technical Guide: Private Voice Transcription with DictaWiz

Practical Summary and Prerequisites

DictaWiz is an agentic AI keyboard and notes manager designed to perform voice-to-text transcription and AI processing entirely on-device. By utilizing local AI models, the application ensures that voice data does not leave the user's phone, providing a private alternative to cloud-based transcription services. The tool functions as both a system-wide keyboard and a standalone notes application, allowing users to dictate directly into other apps or record meetings and lectures for later processing. To utilize these features, users must be running the application on compatible hardware, specifically requiring iOS 26.0 or later, iPadOS 26.0 or later, or macOS 15.6 or later on devices with an Apple M1 chip or later.

The application is designed for users who prioritize data sovereignty, such as legal professionals, journalists, and students who require secure, offline transcription. Because the software operates locally, it functions in environments without internet connectivity, such as airplanes or tunnels. The system is built to handle complex tasks like diarization and summarization without requiring an account, tracking, or external analytics, provided the user does not opt into cloud-based features.

Setup and Local Processing Configuration

The core value proposition of DictaWiz is its ability to operate without an internet connection. Upon installation, the application is configured to run all AI models locally by default. Users do not need to create an account or provide tracking data to begin using the core transcription features. The application supports 25 languages via a fast engine and over 99 languages using a comprehensive engine, with auto-detection capabilities available. Because the system is designed for privacy, no data is transmitted to external servers unless the user explicitly chooses to connect their own API key for optional cloud-based features, which is never a requirement for standard operation.

To maintain the integrity of the local-first workflow, users should ensure that they do not input personal API keys unless they specifically intend to utilize cloud-based transcription, speaker identification, or LLM features. The app provides a robust dictionary and vocabulary management system that learns from user input, allowing for the correction of names and technical jargon that the AI might otherwise misinterpret. This auto-learning dictionary ensures that the transcription quality improves over time as the user interacts with the system, all while keeping the vocabulary data stored locally on the device.

Documented Workflow for Transcription

The workflow for capturing voice notes is flexible, supporting both real-time dictation and file-based processing. For real-time needs, users can integrate the DictaWiz keyboard into any application, such as Messages or Mail, to convert speech to text instantly. The app also provides a dedicated recording interface for longer sessions like meetings or interviews. During these sessions, the application performs diarization to label speakers and generates automated summaries, including key points and action items. Users can also import existing audio files or URLs for transcription. The application includes features to manage these notes, such as creating flashcards, quizzes, or mind maps from the generated transcripts.

For users who require specific formatting, the application offers a Canvas feature that turns transcripts into editable documents with lists, callouts, and key points. This allows for the post-processing of raw transcripts into structured study material or professional notes. The app also supports the use of abbreviations that expand as the user types, which is particularly useful for medical or legal professionals who frequently use specialized terminology. By setting up these snippets, users can ensure that their dictated text is accurate and formatted correctly from the moment it is transcribed.

Advanced AI Skills and Automation

DictaWiz includes an agentic AI keyboard that allows users to execute complex tasks through built-in shortcuts or custom-defined skills. These skills can be chained together to trigger sequences of actions automatically after a recording is completed. For example, a user can dictate a note and have the system automatically format the text, extract action items, and export them to other applications. The application supports integration with various productivity tools, including Obsidian, Notion, Bear, and Drafts, or any service that accepts webhooks. Users can also define custom vocabulary to improve transcription accuracy for specialized jargon, medical terms, or legal phrases.

The automation capabilities extend to the keyboard itself, where users can tap AI Skills to perform tasks such as fixing grammar, inserting map links, adding calendar events, or drafting replies. If a specific skill is not available, users can describe the desired functionality in plain English, and the application will build the skill. This agentic approach allows for a highly personalized workflow where the application adapts to the user's specific needs, whether they are managing a project, studying for an exam, or conducting professional interviews.

Limitations and Data Privacy Considerations

While the application is designed for privacy, users should be aware of the technical constraints inherent in on-device processing. The performance of transcription and AI summarization is dependent on the hardware capabilities of the host device. Although the app is designed to work offline, certain advanced features or integrations may require network access if the user chooses to connect external services. The developer, CORDOC LLC, notes that privacy practices may vary based on the specific features enabled. Users should review the official privacy policy at https://freevoicereader.com/privacy to understand how data is handled, particularly regarding identifiers used for analytics if the app is not used in its strictly offline configuration.

Furthermore, users should be mindful of Apple's restrictions on third-party keyboards, which may occasionally result in minor performance issues or rough edges. The developer has indicated that they are actively working to address these limitations through regular updates. Users are encouraged to report bugs directly to the developer to assist in the ongoing improvement of the application's reliability and performance. By maintaining a clear understanding of the distinction between local processing and optional cloud-connected features, users can effectively leverage the tool while maintaining their desired level of data privacy.